KENWOOD TS-940 PAGE

Back to Part 1 of TS-940 page: Click here

4.4 Experience from Persons who modified the TS-940. 2

4.5 : Pin Diode Modification for TS-440. 2

4.6: So summary of Pin Diode Modification. 2

RECEIVER 5: S METER SLAM AND IMPROVED PERFORMANCE. 2

RECEIVER 6: PHASE NOISE. 2

6.1 Observation of the Phase Noise problem by Thomas Hohlfeld DF5KF in 2005. 2

6.2 Extensive History of the Phase Noise Issues upto about June 1987. 2

6.3 Data Sheet for Problem BA718 OpAmp. 2

6.4 What is Phase Noise and how to measure it by John Grebenkemper, KI6WX.. 2

6.5 Phase Noise measurements on Elecraft K2 by John Grebenkemper, KI6WX.. 2

6.6 Significant Phase noise improvements developed by Thomas Hohlfeld DF5KF.. 2

6.7 Information on Ultra Low Noise OpAmps available in 2008. 2

6.8 Further assessment and Phase noise improvement evaluations planned by Jeff King ZL4AI 2

6.9 First plots of phase noise comparisons below show some very exciting results for DF5KF 940: 2

7.0 SM-220 mods to remove Ghost Signals generated by the 940. 2

PLL BOARD PROBLEMS.. 2

PLL BOARD 0: Reseat Connectors. 2

PLL BOARD 1: Remove the Black Foam from Behind the Board. 2

PLL BOARD 2: Remove the Wax from the VCOs. 2

PLL BOARD 3: Identify which PLL is not locked. 2

PLL Board 4: PLL Board and RF Board and PLL out of lock. 2

PLL Board 5: PLL Board and setting voltages. 2

CONTROL BOARD.. 2

VOLTAGE REGULATOR HEATS UP AND CAUSES A SHIFT IN BFO ON IF BOARD.. 2

AVR BOARD & POWER SUPPLY.. 2

FAN AND TEMPERATURES.. 2

COOL AVR COMPONENTS BY REMOUNTING ON HEAT SINK.. 2

Power Supply HEAT SINK RUNS TOO HOT.. 2

VERIFY THERMISTOR 101 IS ATTACHED AND FUNCTIONING.. 2

REPLACE Q101 AND Q 102: THE MOST DANGEROUS DEFECT OF THE 940.. 2

SAFETY PROCEDURES WHEN Q101 AND Q 102 HAVE FAILED:.. 2

MOTOR BEARINGS GUMMED UP: TEMPORARY FIX.. 2

VK5KYO includes Larger Computer Fan and relocated Rectifiers. 2

28 Volt Crowbar safety circuit for the TS-940. 2

4.4 Experience from Persons who modified the TS-940

First from www.contesting.com

TopBand: : [WSVHF] TS-940 Specs

km1h @ juno.com mailto:km1h@juno.com
Thu, 05 Mar 1998 10:30:41 EST

To all those on Topband who asked...here is the results of the TS-940 tests.

73  Carl   KM1H

--------- Begin forwarded message ----------

From: km1h@juno.com (km1h @ juno.com)

To: wsvhf@qth.net

Subject: [WSVHF] TS-940 Specs

Date: Tue, 09 Sep 1997 22:35:15 EDT

Message-ID: <19970909.214104.9687.15.km1h@juno.com>

As an addition to the Sherwood receiver info,  here are some specs on a TS-940.

All tests were run on 28MHz during the past few days on a customers unit.

TS -940 late serial #  with factory phase noise updates:

MDS SSB  -135 dBm

MDS CW cascaded 500Hz filters   -137 dBm

Sensitivity 10dB S+N/N   .12uv

Phase Noise    -131dBc   at 10 KHz

Filter rejection CW  >90 dB

2 tone dynamic range Wide  95 dB @ 20 KHz

2 tone dynamic range Narrow ( CW Filters)  77 dB @ 2 KHz

3rd Order IP  +1dBm at  20 KHz

Wide Band transmitted noise  -75 dB below full carrier

The same TS-940 but with PIN diode mods to RF and IF boards:

MDS SSB  -137dBm

MDS CW   -142 dBm

Sensitivity  ~ .1uv

Phase noise  -131 dBc  @  10 KHz

Filter rejection ( CW )  >90 dB

2 tone Dynamic Range Wide  102 dB @ 20 KHz

2 tone Dynamic range Narrow  ( CW Filters)   83 dB @ 2 KHz

3rd Order IP   +5 dBM @ 20 KHz

Wide band transmitted noise  -90 dB below full carrier

A few notes and comments:

Although the PIN diode improvement is evident in the numbers the audible difference is much greater.

First of all the receiver is noticeably quieter. The IMD performance shows an "apparent" improvement of

about 10-12 dB under crowded band conditions. This follows along with conversations I had with Dr. Ulrich

Rhode several years ago when I first started using PIN diodes. The cumulative effect of multiple strong signals degraded IMD performance in a stock receiver a lot more than a simple 2 tone test would indicate.

The receiver is a pleasure to use in lowband pileups now.

The improvement in wideband TX noise is due to, I believe, the use of PIN's in those paths that were

common to TX and RX on the IF board. The stock diodes either generated noise or allowed RX path noise

into the TX path. The same appears to hold true in the opposite scenario. This TX noise is something I

was recently made aware of by a local on 6M and bears closer examination and possible additional

improvements.  The noise does not change dB levels when going from full power to the 20-30mw from the

transverter port so it can not be blamed on thermal noise in the subsequent linear amp stages.

Magazine reviews of the TS-940 were of early production. Kenwood at first refused to admit to phase noise

problems. They then went thru two different mods before they were satisfied. The improvement between early

and late models is about 15 dB.

Serial numbers in the mid 8 Million group  and up had factory mods.

Kenwood Service Bulletin 917 may be retrofitted to the earlier radios. It

is a fairly simple mod.

Other KW radios such as the TS-850 and TS-930 could also benefit from PIN mods. They both have transverter

ports and are quite reasonably priced on the used market.

I cant speak for other brands but a quick review of a few Service Manuals shows an awful lot of commonality

across all brands with respect to diode switching schemes.

73....Carl   KM1H

----- Submissions: wsvhf@qth.net Subscription/removal:

wsvhf-request@qth.net

--------- End forwarded message ----------

====================================================================================

-----Original Message-----
From: thomas hohlfeld [mailto:thomas_hohlfeld@hotmail.com]
Sent: Wednesday, 27 April 2005 10:42 a.m.
To: jaking@es.co.nz
Subject: RE: ts-940

Hi Jeff,

thank you for your reply and congrats for the informative web site. I read about the PIN diode mod in a web message.

I replaced diodes D9 through D20 of the RF board by the PIN types BA479. I did not replace D3-D8 because these

switch a frequency too low for the BA478 to be an effective PIN. I checked the BA478 and found them to be good at frequencies above 2 to 3 MHz. The BA479 have been offered in different versions and it is important to use those

which are designed for HF (not VHF). Certainly, other PIN diodes may also be useful. The IMD of my 940 improved by

about 5 dBm after this mod.

For re-alignment of the receiver, I mainly followed the instructions of the service manual for the RF and IF

amplifier stages (I did not align oscillator and PLL circuits). I have a sweep generator (Rohde & Schwarz SWOB 5

equipped with log amplifier), which was very helpful to optimise the bandpass filters on the RF board.

I established RX sensitivity with a HP8640B RF generator together with a home made audio voltage detector to

determine an audio increase by 10 dB. I have two of the HP8640B, so that I am able to determine receiver IMD.

The HP8640B are quite famous, since they produce a very clean RF signal and are sold at a reasonable price.

By the way, I own two TS940 and use one of them for experimental modifications which are more ´critical´, so

that I would not really recommend others to reproduce them. If you are interested anyway, I will report on that

later. It’s past midnight now.

73 for today,

Thomas

>From: "Jeff King" <jaking@es.co.nz>

>Reply-To: <jaking@es.co.nz>

>To: <thomas_hohlfeld@hotmail.com>

>Subject: RE: [ts-940] Re: Why don't more people use this group?

>Date: Mon, 25 Apr 2005 09:09:42 +1200

>Thomas

>I found your review of the FETS most interesting.

>My reason for writing is to ask you to tell me more about the pin diode modifications you have undertaken.

>Where are these diodes and what do you do to replace them?

>What else did you do to realign receive?

>How did you establish the 0.15 uV sensitivity.?

>Yours sincerely

>Jeff King ZL4AI

==========================================================================================

>From: "Jeff King" <jaking@es.co.nz>

>Reply-To: <jaking@es.co.nz>

>To: <jaking@es.co.nz>

>Subject: RE: ts-940

>Date: Fri, 29 Apr 2005 17:05:30 +1200

>Hi Tom,

>Thanks this information is very interesting.

>I really appreciate your advice on aligning the 940 very helpful information.

>I also read [below] about pin diodes being changed in the IF board RX / TX circuits. Have you tried changing any of those?

>Is the Temic's- Vishay (former Telefunken) BA479 (G or S suffix) a suitable diode? Spec sheet attached.

>Yours sincerely

>Jeff King ZL4AI

-----Original Message-----
From: thomas hohlfeld [mailto:thomas_hohlfeld@hotmail.com]
Sent: Monday, 2 May 2005 9:23 a.m.
To: jaking@es.co.nz
Subject: RE: ts-940

Hi Jeff,

Thank you for the comments on the pin diodes. Yes, I used the BA479G. These were quite inexpensive and it was

easy to obtain them when I did this mod two years ago. Today, I checked for a source in the internet

and found a German distributor who may still have the BA479G. Check at

http://www.schuricht.de/w3a/default.asp (a button at the lower left will switch language into English).

They do not look too expensive (0.31 Euro when 50

are ordered).I checked the BA479G with an RF generator and scope and found it to be good at 7 MHz and

higher. I am not sure if it is very effective at 3.5. Nevertheless, I decided to replace D7 through D20

on the RF board. I think I also exchanged D21 (don’t know any more, but makes sense). I did not change

D23 and D26 in the preamplifier because these probably are already pin diodes and their exchange might

impair the AGC characteristics. There are some switching diodes also in the IF section of the 940, but I

did not replace these, because XF-1 already cuts the bandwidth at the front of the IF amplifier and

there may not be too much intermodulation behind. May be I will do this later when I have to take out the

IF board for some other reason.

I told in my last mail that I did other mods in my 940s. It may take quite long to describe them all, so

for the first time here is a short list:

(1) I exchanged the PLL amplifier IC (IC18, PLL board) into a pair of extremely low noise OPamps,

which lowered phase noise. However, this required to tackle some problems with PLL instability.

(2) I have the optional 250 Hz CW filter for the 455 kHz IF. When working PSK in SSB mode, I missed the

possibility to activate this filter. I found out how to modify the 940 to allow for activation of the

CW filter in SSB, including the control LED at the narrow CW filter switch.

(3) When I bought my first 940, it came without the AT unit. So I built one with an automatic antenna tuning

board (kit) and built an interface which nicely communicated with the antenna tuning control circuit of

the 940. Later, I got the original AT-940.

(4) Follow hyperlink

VOLTAGE_REGULATOR_HEAT_UP_AND_BFO_SHIFT

(5) I equipped my 940s with the piexx boards which allows to control the 940 via the serial interface of

a PC.

Best 73s for today, Jeff

Thomas, DF5KF

==========================================================================

The following 2 emails have had some irrelevant content edited out.

>From: "Jeff King" <jaking@es.co.nz>

>Reply-To: <jaking@es.co.nz>

>To: "'thomas hohlfeld'" <thomas_hohlfeld@hotmail.com>

>Subject: RE: ts-940

>Date: Fri, 27 May 2005 19:35:16 +1200

>Hi Thomas,

>Thanks for your email which I am still thinking about, a lot.

>Anyway I have been trying to get BA479G diodes from

>a supplier.

>The supplier advises they have the BA479 but cannot tell if it is a G or S.

>I have asked them to put a resistance meter on it and await those

>results. [They replied they could not help.]

>Attached you will find the datasheet. As you can see on the second

>and third lines on the first page

>Reverse impedances are:

>G: 5 k Ohms

>S: 9 K Ohms

>It appears to me that the S might be better, or no worse than the G.

>Do you think S would be suitable to use without further testing.

>My worry is installing an S and finding it does not work well.

>I have an interesting article from Radcoms Pat Hawker on Pin Diode

>replacements in July 1995. If you would like this I will send that when

>I have it scanned. This article explains that the BA482 replaced in the

>Omni VI with the HP 5032-3081 resulted in impaired performance because

>the HP put through 0.5 dB less signal.

>Your sincerely

>Jeff King

-----Original Message-----

From: thomas hohlfeld [mailto:thomas_hohlfeld@hotmail.com]

Sent: Tuesday, 31 May 2005 9:15 a.m.

To: jaking@es.co.nz

Subject: BA479 etc.

Hi Jeff,

Thank you for your mail and your thoughts on the PIN diode mod. Yes, you are

perfectly right in that this mod will probably be a tradeoff between

linearity and a slight loss of sensitivity. I would be very interested to

read Hawker´s article you mention.

Your mail also made me re-think about this mod. The G version of the BA479

has been suggested to be better for short wave purposes, but it is true that

there is not much evidence in the datasheet. My guess is that the BA479

should simply be measured in a test setup, and this is what I started

yesterday.

I set up a simple circuit with two RF generators (300 mV out) fed into a

hybrid combiner, the output of the combiner going into one end of the diode

under test. The other end of the diode was coupled via a step attenuator

into a spectrum analyser (my shack is a museum of old instruments, hi). I

fed a forward current of about 10mA through the diode, uncoupled from RF of

course. The generators were adjusted between 2 and 15 MHz, always 500 kHz

apart (two tone signal).

I hadn’t much time and did only some very short measurements. Anyway, the

results were interesting. Here they are:

1. A conventional all-purpose diode (1N4148) produced a horrible spectrum of

intermodulation products. In addition, the insertion loss was high (up to 10

dB).

2. Next, I inserted one of the original diodes which I replaced in my TS-940

(of course I did not through them away). The difference was impressing. At

10 MHz and above, these diodes produced very little intermodulation

distortion with very low insertion loss (1 dB or so). Below 10 MHz, the

intermodulation became worse and was poor at 3.5 and below. Interestingly,

insertion loss moderately increased below 7 MHz (reaching 3 dB at 3.5).

3. I also tried a BA479G which I left from another project. These produced

very low intermodulation signals, even below 2 MHz (which surprised me). At

10 MHz and above, the BA479´s have a slightly higher insertion loss than the

original Kenwood diode (1-2 dB worse). Below 10MHz the BA479 showed less (!)

insertion loss and were clearly superior with respect to intermodulation.

To summarize, the BA479G is better than the original TS-940 diode only at

frequencies lower than 10 MHz. As you will imagine, I am thinking about

returning to the original diodes at 10 MHz and above. It may take a couple

of weeks until I will have time.

I can also take some digital photographs of the intermodulation spectra and

mail them to you, if you are interested (may take 2-3 weeks). Let me know if

your e-mail server has limitations in file size.

Best regards for today and vy 73,

Thomas,

DF5KF

-----Original Message-----
From: thomas hohlfeld [mailto:thomas_hohlfeld@hotmail.com]
Sent: Friday, 10 June 2005 8:56 a.m.
To: jaking@es.co.nz
Subject: RE: BA479 etc.

Hi Jeff,

thank you very much for including these excellent articles on the web page.

It’s so useful that I’ll print it out.

Now, here is part #1 of my measurements on the RF properties of the 940 Front end diodes. To start, I concentrated

on the diodes in front of the filters (diodes # 3, 5, 7 ... 19), which see the whole RF spectrum from by the antenna.

The forward current of these diodes is 25 mA in my 940, which appears quite a lot, but probably was chosen because

the diodes need to be very linear here.

The attached file (25mA-all.jpg) shows the schematic test setup on top. I already described it in my previous mail,

but this time I measured at a ´realistic´ diode current of 25 mA. The two RF generators were set close to 7, 3.5 and

1.75 MHz, always 250 kHz apart. The spectra (figs 1-12) show the two carriers at the center. All additional peaks are

3rd and higher order products, indicating the non-linearity of the diodes. To get an ´ideal´ reference, I also

measured with the diode replaced by a wire bridge (figs 1, 4 and 8). Here, the low remaining 3rd order signals

(less than -60dB) reflect some non-linearity of my test setup (my homemade hybrid combiner may not be perfect).

As I already noticed in my earlier mail, a general purpose diode (1N4148) produced a horrible spectrum of side

products at all three bands (figs 2, 6, 10). The original Kenwood diode (1S2588) was very good at 7 MHz (and at

higher QRG´s, not shown), but worse at 3.5 and quite poor at 1.75 MHz (figs 3, 7, 11). In contrast, the BA479G

was very good and provided the least side products (figs 4,8,12). If you look closer at the distance of the carriers

from the top graticule (sorry, background is quite dark), you will notice that the carrier attenuation of the BA479 is

low at all bands and not much different from the original Kenwood diode.

Part #2 of my measurements will follow, where I studied the diodes behind the input filters (D4,6,8...20 and D21)

which are run at a lower forward current. The measurements are already finished but the figures need to be arranged.

Good luck for today!

Thomas

DF5KF

-----Original Message-----
From: thomas hohlfeld [mailto:thomas_hohlfeld@hotmail.com]
Sent: Monday, 13 June 2005 4:31 a.m.
To: jaking@es.co.nz
Cc: hohlfeld@uni-duesseldorf.de
Subject: TS-940S - pin diode part#2 and some other considerations

Hi Jeff,

here comes part #2 of the pin diode measurements, which concentrates on the

diodes between the bandpass and the preamplifier of the TS-940.

First a short comment on the 940´s front-end: when the RF has passed the RX

bandpass, three diodes follow before the signal reaches the preamplifier

Q10. Of these, two are switching diodes (one of D6,8 ... D20, D21), while

one (D26) is a pin diode. Only the two switching diodes are candidates for

replacement. For test purposes, I assumed that D26 is fully open and D23

closed (AGC at highest sensitivity). In this case, we deal with two

switching diodes in series. They are also in series with respect to their

forward current, which is determined by R31. I measured 17 mA in my TS-940.

The impedance, which these diodes see, is also important. I estimate it

about 50 Ohm, because the bandpass filters of the TS-940 are constructed

symmetrically.

Therefore, my test setup was similar as before with the following changes:

(1) I tested two diodes in series, (2) I used a forward current of 17mA and

(3) I also measured at 10.1 MHz. Because the 1N4148 was so poor in part#1, I

did not consider it further. The attached jpg file again contains a plot of

the test setup.

Here are the results (see attached jpg file): At 10.1 MHz (figs 1-3), the

original Kenwood switching diode (1S2588) is as good as the pin diode BA479.

The same also applies to higher frequencies, which I do not show here. The

insertion loss of the two BA479 in series is slightly higher than that of

two Kenwood diodes, but is still less than –1 dB (more on this below). At 7

MHz (figs 4-6), the 3rd order products are clearly increased by the Kenwood

diodes (fig 5) in comparison with the reference (diodes shorted, fig 4). The

BA479 (fig 6) is clearly better than the Kenwood diode, although a minimal

increase is also seen compared with the reference. At 3.5 MHz the Kenwood

diodes generate a lot of intermodulation products (fig 8) and the BA479 (fig

9) is obviously superior. The same is true for the 1.75 MHz band (figs

10-12). In summary, the BA479 is better than the original Kenwood switching

diode at 7MHz and the lower bands at a forward current of 17mA. There is no

relevant difference at 10 MHz and higher.

It would also be interesting to know how the diodes behave at frequencies

below 1.75 MHz. Unfortunately, my combiner is not appropriate for a lower

QRG. But I compared the generation of harmonics by the Kenwood diode and the

BA479 at lower frequencies and found the BA479 still to be better than the

Kenwood diode even at frequencies down to 100KHz.

The above measurements also show that the BA479 has a slightly higher

insertion loss than the Kenwood diode. This is a well known disadvantage of

pin diodes (also adressed in the excellent articles on your web page). In

the case of the BA479, however, the effect is less than –1dB and therefore

probably negligable in the 50 Ohm system of the TS-940. To confirm this, I

also used a dB meter to determine the insertion loss of two Kenwood diodes

in series compared with two BA479 in series. At 17 mA diode current, two

Kenwood diodes produce a loss of -0.2dB at 1.75 and -0.1 dB at 3.5 though 28

MHz. Two BA479 in series cause a loss of –0.8 dB at 1.75 and 3.5MHz and

-0.7dB at 7 through 28MHz. So there is a clear difference, but probably

without much importance.

What are the consequences? As long as a broadband antenna feeds the TS-940,

the front-end before the bandpass (diodes D3, D5... D19) will probably be

improved by changing into suitable pin diodes, such as the BA479. Those who

use the TS-940 only with a beam antenna (e.g. 20-10m), which is unlikely to

deliver large signals at 80 and 160m, will probably not have much benefit.

The two switching diodes between the bandpass filters and the preamplifier

may also be replaced by pin diodes at the lower bands (D6, D8 ... D14 and

D21). I would not recommend to exchange D16, D18 and D20, because the

original diodes are already excellent at the higher bands and the pin diodes

would add nothing else than a (minimal) increase of insertion loss.

In addition to the front-end diodes, there are also numerous switching diodes

in the IF unit. Particularly those before the 8MHz crystal filters may be

considered for replacement by pin diodes. I can imagine that this will

improve narrow-band intermodulation. I’d be curious if anybody else has

experience with this. If not, I may check out this point in future.

---------------

Your last mail says you ordered 40 BA479 diodes, so it seems you found a

source. I for myself also ordered 50 BA479 from Schuricht, here in Germany.

It’s good to have some on stock for future projects and people say these

parts are likely to be replaced by SMD types in future. If you still need

BA479, let me know and I will try to help. By the way, there are probably

excellent alternatives. The TS-940 uses in its front-end an attenuation pad

which is part of the AGC (D23 and 26, MI204). I tested these and found they

are even a little better than the BA479. The problem with these is that they

are hard to obtain.

---------------

Thank you for mailing the discussion on the reversed Q10/Q4 problem.

Although I was unable to find a difference in my test setup (as I reported

earlier), it is certainly possible that there is an asymmetry of the

internal capacitances that did not become apparent in my measurements. I

think I should try out this mod and do some measurements, but it may take a

little until I’ll have time. It would also be interesting to see how the

increase in sensitivity, if it really occurs, will change the receiver’s

dynamic range.

---------------

Finally, my congratulations for your exciting web page! You did a very good

job in digging out all the fascinating information about and around the

TS-940. It’s a pleasure to contribute.

Best regards,

Thomas

(DF5KF)

4.5 : Pin Diode Modification for TS-440

A similar sort of mod for the TS-440 maybe found at:

http://www.mods.dk/view.php?ArticleId=1709

===========================================================================

4.6: So summary of Pin Diode Modification

Until further research verifies otherwise, only replace (odd numbered) diodes that carry less than 10 MHz. [Above 10MHz BA479 has more attenuation than the Kenwood 1s2588, so leave the original 1s2588 in place.] It maybe better to hold off any replacement until research is completed.

-----Original Message-----
From: Jeff King [mailto:jaking@es.co.nz]
Sent: Monday, 13 June 2005 9:53 p.m.
To: 'thomas hohlfeld'
Subject: RE: TS-940S - pin diode part#2 and some other considerations

Hi Thomas,

Thanks for this.

Just to verify I understand you correctly.

At 17 ma you advise

=============================================================

The above measurements also show that the BA479 has a slightly higher

insertion loss than the Kenwood diode. This is a well known disadvantage of

pin diodes (also addressed in the excellent articles on your web page). In

the case of the BA479, however, the effect is less than –1dB and therefore

probably negligible in the 50 Ohm system of the TS-940. To confirm this, I

also used a dB meter to determine the insertion loss of two Kenwood diodes

in series compared with two BA479 in series. At 17 mA diode current, two

Kenwood diodes produce a loss of -0.2dB at 1.75 and -0.1 dB at 3.5 though 28

MHz. Two BA479 in series cause a loss of –0.8 dB at 1.75 and 3.5MHz and

-0.7dB at 7 through 28MHz.

=================================================================

so Jeff King concludes that

Two BA479s at 17 ma show

-0.6 dB more insertion loss at 1.75 MHz

-0.7 dB more insertion loss at 3.5 MHz

-0.6 dB more insertion loss at 4 MHz through to 28 MHz

This means the BA479 will reduce the signal and therefore will not improve the receiver performance. In Jeff’s opinion this is considerable increase in loss of receiver performance, and not really viable to use a BA479 as a replacement.

Transmitter performance between 1.5 and 7 MHz will improve because of the much better BA479 IMD performance you describe.

On 31-5-2005 Thomas advised at 25 ma

==============================================================

3. I also tried a BA479G which I left from another project. ……… At

10 MHz and above, the BA479´s have a slightly higher insertion loss than the

original Kenwood diode (1-2 dB worse). Below 10MHz the BA479 showed less (!)

insertion loss and were clearly superior with respect to intermodulation.

==============================================================

This means the only diodes that should be replaced are

D13, D11, D9, D7. in operating at 1.5 MHz to 8.5 MHz and 25 ma

D5 and D3 operate at less then 1.5 MHz, and the performance of the BA479 is unknown in that region.

I wonder if you could document the actual insertion loss of a BA479 performance at 25 ma between 1.75 MHz AND 10.0 MHz?

I look forward to your reply.

Yours sincerely

Jeff King

Diodes on RF board	Original Kenwood Diode	Original Spec	Operating Frequency MHz	Replacement	Replacement Spec
D3	1s2588 [(L30) DIODE TW-4000A, $3.15]		~0.5	Don’t replace until testing verifies suitability
D4	1s2588		~0.5	Don’t replace BA479 -0.3 dB loss is too great
D5	1s2588		0.5-1.5	Don’t replace until testing verifies suitability
D6	1s2588		0.5-1.5	Don’t replace BA479 -0.3 dB loss is too great

D8	1s2588		1.5 -> 3.0	Don’t replace BA479 -0.3 dB loss is too great
D10	1s2588		3 - 4	Don’t replace BA479 -0.35 dB loss is too great

D12	1s2588		4 - 7	Don’t replace BA479 -0.3 dB loss is too great
D14	1s2588		7 - 8.5	Don’t replace BA479 -0.3 dB loss is too great
D7	1s2588		1.5 -> 3.0	PIN type BA479	SI-D 30V 50mA 100MHz
D9	1s2588		3 - 4	PIN type BA479	SI-D 30V 50mA 100MHz
D11	1s2588		4 - 7	PIN type BA479	SI-D 30V 50mA 100MHz
D13	1s2588		7 - 8.5	PIN type BA479	SI-D 30V 50mA 100MHz

D16, 15	1s2588		8.5 - 14	don’t replace BA479 has 1 - 2 dB higher insertion loss than 1s2588 > 10MHz
D18, 17	1s2588		14 - 20	don’t replace BA479 has 1 - 2 dB higher insertion loss than 1s2588> 10MHz
D20, D19	1s2588		2 -- 30	don’t replace BA479 has 1 - 2 dB higher insertion loss than 1s2588> 10MHz
D21	1s2588		2 -- 30	don’t replace BA479 has 1 - 2 dB higher insertion loss than 1s2588> 10MHz
D26	MI204	Pin Diode	2 -- 30	don’t replace BA479 has 1 - 2 dB higher insertion loss than 1s2588> 10MHz

Whether you install many pin diodes depends on how crowded the bands are at your location.

More information below explains this:

-----Original Message-----

From: thomas hohlfeld [mailto:thomas_hohlfeld@hotmail.com]

Sent: Tuesday, 21 June 2005 9:16 a.m.

To: jaking@es.co.nz

Subject: RE:

Hi Jeff,

yes, I think that 0.3-0.4 dB is almost nothing. Remember that one S unit makes 6 dB (20-fold more!). The additional loss caused by the pin diodes therefore is a very small fraction of an S unit. A few meters of RG58 antenna cable with conventional plugs cause more attenuation than these pin diodes.

By the way, I also measured the attenuation between the TS-940 antenna plug and the preamplifier input. All frontend switching diodes plus the relay contacts plus the bandpass filters (14 MHz) have an attenuation of 2 dB. The same measurement with my Icom IC751 was 5 to 6 dB. This shows that the TS940 is indeed an excellent construction.

Based on the results of my pin diode measurements I have now replaced most of the switching diodes with BA479G. The only exception are D20, D18, D16 and D14 which I did not change. With this change (plus turning around Q10), the RX sensitivity (MDS) is -135 dBm, an excellent result fully competitive with the top Rigs marketed today.

Best regards,

Thomas

From: Traian Belinas [mailto:traian.belinas@deck.ro]

Sent: Monday, 20 June 2005 10:29 p.m.
To: jaking@es.co.nz
Subject: Re: pin diode, reversed FET

Hello Jeff,

0,35 dB is an INSIGNIFICANT amount of loss, it is even dificult to detect a such low difference....
The positive aspect of the reduced IMD (especially the second order IMD reduction) is by far more important than the little loss.
My only doubt in such case is the parts and labour cost, and if the mod is really needed for you there.
If you consider it as being needed (if you have unwanted strong signals at your QTH, and please consider the broadcast bands also), then go for this mod with confidence.

Thomas is right, the lower bands are really crowded here, especially during the evening. You are a lucky OM being there in ZL...
After 22.00 local time, the 80m band is full here, I can hear even the italian and DL stations making local QSO's with other I and DL stations there respectively and having some big signals, as not to mention the russians which are everywhere and really strong, like locals, and S9 + 30 to 40 dB signals are usual. The thousands of GU43, GU74 and GS35 power tubes are really put to work out there, hi.
In my case, the added city QRN is also high, the normal noise is to +20 dB, so I use the attenuator for to get the noise lower, at a reasonable level as it have no sense seeing the S meter to S9 + 20 only because of the noise and so to loose a big part of available Rx dynamic range. The PIN mods are usefull here in Eu.

73,
Traian, YO9FZS

All the Best,
Traian

From: Traian Belinas [mailto:traian.belinas@deck.ro]
Sent: Wednesday, 22 June 2005 10:26 p.m.
To: jaking@es.co.nz
Subject: Re: pin diode, reversed FET

Jeff King wrote:

Traian,

Hi thanks for this. It did really help my understanding.

In New Zealand we have max of 4,500 hams.

The bands are not crowded

http://kb9amg.slyip.com/markd/KB9AMG/top_dx_spots/by_callsign/zl.html

This is the strongest ZL stations.

So on 80 meters at night I can tune whole band and only hear about 2 or more other stations.

My 80 antenna, you have seen diagram is only about 13.5 off ground so it does not work well.

Usually I hear Australian stations and occasionally some USA.

Yes when propagation is there I hear those strong Russians here too.

Now back to Pin diodes.

My simple understanding of how the pin diode works is that the clean pin diode prevents other unwanted multiple frequencies up and down the band. This would seem to be an advantage during transmitting because it prevents unwanted additional off frequency splatter signals on the band.

Until I read your email yesterday, I did not comprehend to the fact that of course this works in reverse for receive!!! It the existing diodes generate those off frequency signals then of course the a nearby signal will be picked up off frequency in reverse the same way and that off frequency signal will be heard as interference on top the signal you want to hear. Hence Pin Diodes significantly improve the selectivity of the receiver. Traian please confirm this understanding is correct.

Not quite so.
The PINs will have big advantage for Rx only.
The problem with the normal junction diodes is that when they conduct (when the filter is switched in line), the far out of band signals may cause 2nd order IMD and the close or in band signals may cause 3rd order IMD. Note that the in band mean inside the BPF which may have many MHz bandwidth...
Actually the diode act like a mixer (such simple one diode mixers are used, especially at SHF)! You may see how it hapen if consider the unwanted signals which will be mixed by the diode...
For example, when receiving the 14 MHz band here in EU, if big signals exist in the 41m broadcast band, they may cause second order IMD (a 7110 and 7150 kHz BC stations may produce a ghost signal at 14250 kHz as second order IMD; the same will hapen for different bands/frequencies...). Also, when receiving the 40 m ham band, two signals on the 41m BC band may cause 3rd order IMD apearing as ghost unreadable or carrier signals on the 40m amateur band. These are only particular examples, as when the propagation is good, many hundreds of signals arrive simultaneously at the RX diode bandpass filters input which may cause a lot of trouble, especially the stronger ones, the band may seem noisy or ghost or unreadable signals may appear.
The advantage of the PINs is that they act (theoretically) as controlled resistor, they can rectify and mix only for the signals at frequency lower than the one corresponding to its carrier lifetime and their switching characteristics regarding the produced IMD are much better than for the normal diodes.
You may understand now why PINs having large carrier lifetime specification as the BA479 are better for HF than the ones having very small carrier lifetime: they may maintain the same good IMD behaviour at lower frequencies; and the bigest problem is at the lower bands, as Thomas measurements confirm, it is just a practical confirmation of the theory...

Regarding the TX, the signals switched during Tx are few, they are the mixing products from the Tx mixer and IMD is not a problem, as all are originating from the same signal, so a PIN will not make large improvement for Tx. Actually, the Tx IMD are generated by the final amplifier...

Jeff, please note that the proffesional Rx, if not using relays for the BPF switching then they are using good PIN diodes.

So, using PINs instead of the existing diodes is a good thing, especially if the above mentioned problems seemed to occur (so include the presence of the strong signals of the broadcast bands!).
But if big signals are not present, if have a good quiet location like yours, and when considering the cost and the effort involved, it may not worth doing it. It is only a decision of each of us, depending also on the local Rx conditions...
This is what I intended to let you know before.

The mod shall be more usefull for me here, but I am not decided because I will have to sell all the radios some day, so no reason for spending $ and effort for such mod, and I consider also that some buyers don't like buying the modified radios...

Now I must admit I am not sure if I Need these pin diodes. Now propagation is not good I am only really listening on 20 at about 3 UTC. According to Thomas [if one want to avoid losses] I probably should only be interested in changing the diodes below 10 MHz, so this is really only going to improve 40m and 80m

Yes, this is done on some Rx, as the Icom R9000, where the Rx BPF are splitted in two banks (LF+MH and HF) switched by good PINs.
the other diodes are normal.
[The TS950 SDX Rx BPFs are also separated in two banks by a HPF filter which attenuate the LF/MF signals when using the higher bands, for avoiding the 2nd order IMD caused by the LF/MF broadcast signals. This is the only difference between the TS950SD and TS950SDX regarding the RF board/front end , and some amateurs that are not aware of this are still speaking of the "big difference" between the two radios !!!]

Now I’m convinced not to do the PIN diode mod. I think you are right I do not need it here, where there is little interference.

==================================================================

I look forward to your reply.

Yours sincerely

Jeff King

RECEIVER 5: S METER SLAM AND IMPROVED PERFORMANCE

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Martin Sole
Sent: Tuesday, 11 July 2006 2:31 p.m.
To: ts-940@yahoogroups.com
Subject: [ts-940] S meter slam mod

I saw a note here the other day/week about the S meter pinning over on power down and an unpublished fix.
I have a radio here that is doing just that, what was the fix? Cannot see it anywhere in the various references checked so far.
Thanks
Martin HS0ZED

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Jeff King
Sent: Wednesday, 12 July 2006 9:53 a.m.
To: ts-940@yahoogroups.com; kenwood@mailman.qth.net
Subject: RE: [ts-940] Re: S meter slam mod: A Fix does exist but is not documented out in the public forum.

Garey, Not all 940s make an S meter slam on shutdown. On

http://homepages.ihug.co.nz/~jaking/TS-940_02.htm#KINR_email

KI4NR [a Kenwood repair expert] explains,

One other thing .....on all the older 940 4, 5 and early 6 mil serial number ...the IF board is different. The gain distribution in not the same. All the 940 ... late 6 and newer had better IF boards. They have more gain TX & RX the radio are hotter sensitivity wise, better AGC compression. I use a 5 mil TS 940 with a later model 8 Mil IF board in it .......much , much better !!

Also Kenwood put an S meter slam mod in those boards. all the older 940 when you shut the radio off, pin the S meter over. The newer boards are fix for that.

If someone can get hold of a later than late 6 mil board, that person will find that extra capacitors / diodes / transistors have been patched onto the rear of the board. There may also have been changes on the on-board components.

If someone could fully document all the changes on the late 6 mill or later board then we would all know what to do to improve the receive. Kenwood must have made a factory modification which has not been publicly released. Once the late 6 mill plus board has been documented, one would have to identify the changes that were due to already published service bulletins. The remaining changes are the secret to improving the receiver the way Kenwood did. If someone could access the unpublished Kenwood Service Bulletins for these changes then that would be the ultimate information for any 940 owner.

Yours sincerely

Jeff King ZL4AI,

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Garey Barrell
Sent: Wednesday, 12 July 2006 3:31 p.m.
To: ts-940@yahoogroups.com
Subject: Re: [ts-940] Re: S meter slam mod: A Fix does exist but is not documented out in the public forum.

Hi Jeff -

Well, mine is not only a "mid 7 mil" unit, it's even been to KI4NR for a complete workover, and it still pegs the S-Meter at turn-off.! Always has.
Typical Kenwood, nothin's ever easy. (Or documented!) :-)

Good to hear from you!
73, Garey - K4OAH
Atlanta

Drake 2-B, 4-B & C-Line Service Data
<http://www.k4oah.com>

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of DavidGriffin
Sent: Thursday, 13 July 2006 12:08 a.m.
To: ts-940@yahoogroups.com
Subject: [ts-940] Re: S meter slam mod: A Fix does exist but is not documented out in

Mine does not.

Dave

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Martin Sole
Sent: Thursday, 13 July 2006 1:22 a.m.
To: ts-940@yahoogroups.com
Subject: Re: [ts-940] Re: S meter slam mod: A Fix does exist but is not documented out in

Sooo... the question has to be, what's your serial number, the radio's I
mean. Starting to look like a cutoff in the 7 mil region.

73
Martin, HS0ZED

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Martin Sole
Sent: Thursday, 13 July 2006 2:52 a.m.
To: ts-940@yahoogroups.com
Subject: Re: [ts-940] Re: S meter slam mod: A Fix does exist...

A quick check now I have the 9 mil radio fixed, at least as far as getting the power supply going anyway, shows no tendency to pin the
meter needle on power down in the way my 7 mil radio does. All this one does is drop to zero immediately followed by a slight blip to about S6.

Now to investigate the difference.
73
Martin HS0ZED

Editor Comment: It looks like Martin is going to investigate and solve the mystery of the S Meter Slam / Receive performance improvement. Here is a question asked 2 days earlier:

Jeff King wrote:

Hi Martin,

Thanks for your communication.

…..

Re The S meter slam. If you find the solution to that I would be eager to know. As KI4NR advised Kenwood fixed that from 7 mil or 8 mil on, but Kenwood never published the fix. I emailed Kenwood Australia about this but they refused to give the information. Whatever Kenwood did to the 8 mil board improved reception and finding that out would be so useful.

Yours sincerely

Jeff King ZL4AI

________________________________________

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Martin Sole

Sent: Friday, 14 July 2006 1:22 a.m.

To: ts-940@yahoogroups.com

Subject: Re: [ts-940] Re: S meter slam mod: A Fix does exist...

For those who may have an interest I believe I have identified the cause of and solution for the S meter end stop slam on power down.

Resources were a 7 mil radio with the 'problem' and a 9 mil radio without.

Observation shows a couple of differences in this area, an additional 100u capacitor under the board and a diode in place of a jumper.

Reference TS940 service manual revised edition 1987-3 pages 92 and 93 on page 92 IC1 is located in area H,2. and on the board layout, page 93 at

E,2.

On the 7 mil radio IC1 is fed with 15 volts through R206 and R178. On the 9 mil radio a diode is located between R206 and R178 in the position of J5 on the board layout. Additionally a 100u capacitor is fitted between the junction of this new diode (cathode) and R178 (+ve side) and ground (-ve side).

Overall this gives an increase in the time taken for the supply to IC1 to fall. Increased from approx 300mS to approx 800mS.

I used an ordinary 1N914 type diode and a 100uF 25 volt electrolytic.

73,

Martin, HS0ZED

Jeff King wrote:

Martin,

Thank you so much for investigating and evaluating the effect of the new components which eliminate S meter power off slam. That is very useful

information which many 940 owners will value. I really appreciate your efforts.

Two questions:

On the 9 mill IF board did you identify any other component changes that will effect (make improvement to) the receiver?

Any particular reason you used a 1n914 as opposed to the more commonly available 1n4148?

[With your permission All you findings will be placed on the TS940 page.]

Yours sincerely

Jeff ZL4AI?

-----Original Message-----

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Martin Sole

Sent: Saturday, 15 July 2006 3:53 p.m.

To: ts-940@yahoogroups.com

Subject: Re: [ts-940] Re: S meter slam mod: A Fix NOW exists... are there any 9 MILL IF changes that IMPROVE THE RECEIVER?

Hi Jeff,

So far I have not dug any deeper into the various board differences. I did note a couple of resistors and capacitors under the board of the 9

mil unit that do not appear to be on the 7 mil board.

A couple of things that I have noted is that the 9 mil radio sounds quieter, equally as sensitive but less background noise. It could be alignment, not sure yet. I do know that the noise blanker in the 7 mil radio is vastly better than the 9 mil radio but I put this down to my own alignment. On many of the 940's and 930's that I have seen over the years most have their noise blanker incorrectly set up such that they are usually only partly effective.

Actually I do not know exactly what the diode was that I used for the meter mod, only that it was a small silicon signal diode, could well have been a 1N4148.

Like Garey, K4OAH, I'm not all that sure that the slam is a big problem. Most older radios, Collins, Drake to name but two generally pin their S meters at switch on and also when muted by a connected transmitter, Have never heard of a lot of meter failures. Either way I think it is better engineering to avoid such where possible.

By all means add whatever you want from this to your most useful resource.

73,

Martin HS0ZED

dean hunsinger wrote:

I have a 9 mil s/n radio and at shut down the meter does not hit the > other end just a slight rebound from off position.Is there a freq that this group uses?

________________________________________

From: ts-940@yahoogroups.com [mailto:ts-940@yahoogroups.com] On Behalf Of Martin Sole

Sent: Sunday, 16 July 2006 2:20 p.m.

To: ts-940@yahoogroups.com

Subject: Re: [ts-940] shutdown

Hi Dean,

That sounds about right from what others have said and what I found. Seems to be somewhere in the 7 or 8 mil range that the change got implemented.

73,

Martin HS0ZED

RECEIVER 6: PHASE NOISE

The Phase Noise improvements are probably the most significant improvements possible to the 940.

These will upgrade a 1984 design to a 2008 design.

What follows in this section is:

1. Observation of the Phase Noise problem by Thomas Hohlfeld DF5KF in 2005

2. Extensive History of the Phase Noise Issues upto about June 1987

3. Data Sheet for Problem BA718 OpAmp

4. What is Phase Noise and how to measure it by John Grebenkemper, KI6WX

5. Phase Noise measurements on Elecraft K2 by John Grebenkemper, KI6WX

6. Significant Phase noise improvements developed by Thomas Hohlfeld DF5KF

7. Information on Ultra Low Noise OpAmps available in 2008

8. Further assessment and Phase noise improvement evaluations planned by Jeff King ZL4AI

6.1 Observation of the Phase Noise problem by Thomas Hohlfeld DF5KF in 2005

From: hohlfeld_thomas [mailto:thomas_hohlfeld@hotmail.com]

Sent: Sunday, 3 April 2005 10:08 p.m.

To: ts-940@yahoogroups.com

Subject: [ts-940] VCO phase noise

Hello to everybody

A couple of years ago I came across some test reports on the 940, which claimed that the phase noise of VCO3 is rather poor. I measured

it and, indeed, is was worse than reported for other rigs of the time. A comparison with an IC-751 (my former rig) indeed favoured the ICOM.

Thereafter, I added the PLL modification recommended by Kenwood, which improved phase noise by 5-10 dB. Still, the ICOM is better. Any

suggestions about further mods to improve the PLL circuit (PLL3)?

For example, I can imagine that replacing IC18 on PLL unit by a modern low noise op amp will improve phase noise.

When measuring the phase noise of my 940, I noticed weak unwanted signals about +/- 20 kHz apart from the main signal. Indeed, the rig

receives weak spurious signals about 20 kHz apart from an S9+40 carrier from a high quality RF generator (HP8694B). The spurious

signals (not the main signal) are slightly shifted when the digital display is dimmed. I checked another TS-940 with a similar result. My

conclusion is that the DC-DC unit (which oscillates at about 20 kHz) somehow interacts with the VCO. Attempts to improve uncoupling the

DC-DC unit from the power supply did not help. Did anybody else observe this and, perhaps, could find a solution?

Certainly, the phase noise and weak spurious signals are minor problems. The 940 remains an excellent rig I. But why not thinking about making a good thing better?

Any suggestions are appreciated.

73 de Thomas, DF5KF

6.2 Extensive History of the Phase Noise Issues upto about June 1987

John Delevore in Edition 1 of Low band DXing speaks specifically about his frustration with phase noise in both the 930 and 940.

Link below is 13 pages of very important observations and dissatisfaction with phase noise in the 940 , when first released.

TS-940_InterNational_Radio_users_Supplement_Addendum1Mar1988.tif

Suggest reading these 13 pages carefully.

Key Points are:

Kenwood mod 911 (which ZL4AI has never seen) quoted as AR-81-031, 18 Aug 1986 reduces phase noise by 2 to 3 db

Kenwood mod 917, 2 March 1987, reduced phase noise further to 8 dBc / HZ

Lowe Modifications by Technical Director John Wilson reduced phase noise further to 13 dBc / HZ. Lowe Modifications are introduction of simple loop filters shown at the bottom of page 11. These can be added to the coax connectors as the VC1 and VC2 lines enter the RF Unit.

Points of interest is:

Lowe did the modifications to transceiver 7060088: manufactured in June 1987.

Lowe was UK Kenwood (Trio) agent so would have had in June 1987 Kenwoods latest modification versions including 917.

Why did John Wilson do his tests against a transceiver fitted with only mod 911?

Kenwoods mod 917 is similar to Lowe’s Mod. It is adjusting the filtering of signals in and out of IC18 which is dual OpAmp.

It appears this OpAmp is very noisy.

6.3 Data Sheet for Problem BA718 OpAmp

BA718.pdf

Please note there is no data at all on this sheet about this OpAmp being low noise.

One could conclude this means the OpAmp is noisy.

6.4 What is Phase Noise and how to measure it by John Grebenkemper, KI6WX

6.5 Phase Noise measurements on Elecraft K2 by John Grebenkemper, KI6WX

Well worth reading:

http://home.pacbell.net/johngreb/k2phasenoise.pdf

John shows that phase noise varies per band. So you really need to measure phase noise across all major bands

160, 80, 40, 20, 15, 10 to characterize a receiver performance.

6.6 Significant Phase noise improvements developed by Thomas Hohlfeld DF5KF

http://www.darc.de/g35/df5kf.htm

Thomas has probably done more than any amateur to improve the 940. His work is truly impressive.

This modification is probably the single largest improvement to the 940 receiver in its history

Phase noise of the TS-940

Thomas Hohlfeld, DF5KF

The relatively mediocre phase noise of Kenwoods TS-9 0 is largely produced by the PLL loop of the main VCO. Kenwood developed a modification and published a service bulletin for improvement. Read here how much this modification helps and what else can be done.

--------------

I own a TS-940 for several years and I am still happy with it. Its general performance is competitive with many contemporary high performance radios. Nevertheless, the TS-940 isn’t free of problems. One is the excess noise generated in the RX when there are strong out-of-passband carriers close to the tune frequency. This is caused by noise surrounding the receiver’s local oscillators and, hence, is produce within the receiver. This ´phase noise´ is mixed with receive signals and appears as a noise sideband of the receive carrier. It may not be evident when performing a QSO under standard conditions, but can be disturbing when weak signals are to be detected with strong signals nearby. Local oscillator phase noise has been designated as a main limiting factor of modern

receiver performance [1, 2]. A good discussion of oscillator phase noise and its consequences on system performance is given in [3]. I studied the problem more closely in my TS-9 0 and evaluated some possible cures.

1. Measurement setup for phase noise determination

The measurement setup was simple: I connected my HP8640B RF generator to the TS-940 antenna input. The HP8640B is known for its clean signal, with phase noise below that of most commercial receivers (<140 dBc/Hz [4]).

The TS-940 was switched to CW mode, RF attenuator off, AGC off an RF gain set to maximum. Band width was set to maximum (2, kHz) and the RX tune to 14.100 MHz. The TS-940 audio output was fed into a true RMS voltmeter. A -125 dBm signal from the RF generator produced a defined audio output, which was 10 B above background noise (= audio out with RF generator replace by 50 Ohm termination).

Then, the RF generator was set to 14.200 MHz (+100 kHz carrier offset) an the generator output increased until the audio noise reached the same audio output as the -125 dBm in-passband signal. The measurements were repeated with stepwise reducing the carrier offset. Lower carrier offsets required lower generator outputs to produce the same audio output.

Phase noise was calculated as follows:

Phase noise [dBc/Hz] = -125 dBm (reference) – dBm RF generator output – 33.8 (*)

(*) correction for 1 Hz bandwidth

2. Improvement of phase noise by Kenwood´s recommended modification

The results of all measurements are depicted in figure 1. The grey line represents the phase noise I obtained before any modifications. The phase noise dropped from about -100 dBm/Hz at 4 kHz offset in a log/linear fashion down to -142 dm/Hz with carrier offsets increasing to 50 kHz. There was no relevant change at higher offsets up to 100 kHz.

Next, I installed the Kenwood modification as described in Service Bulletin #917. As shown by the blue line in figure 1, this reduced phase noise by 5 to more than 10 dB, an encouraging result.

Kenwood´s re-design of the PLL apparently moves the loop filter behind IC18 (BA718), which allows for a better suppression of the noise produced by this OpAmp. As a consequence, one would expect less voltage noise FM modulating the VCO varactor diodes. I could not find a detailed datasheet of the BA718, but I assume it is not a particularly low noise OpAmp.

Figure 1: Phase noise versus carrier offset from the tuned frequency (14.1 MHz). The actually measured offset points are marked by black dots. These were later connected by the coloured lines.

All measurements have been performed on the same PLL board and the same rig, using the same measurement setup, but were performed on different days. Offsets <3 Hz could not be measured with the available setup.

3. Replacement of IC18 by an ultra-low noise OpAmp

Noise in PLL systems can be generate by many causes. One is the loop amplifier stage itself. A good discussion of the origin of PLL noise is given in the ARRL Han book [2]. During the past years, ultra low noise OpAmps have become available, which might be more appropriate for a good

PLL design than the BA718. Therefore, I constructed a small replacement board which contains two LT1028 OpAmps, each in an 8-lead SMD (SOIC) package. The LT1028 is a little more expensive than conventional OpAmps, but still affordable. Figure 2 shows the layout of the PCB, the complete board and part of the TS-9 0 PLL unit after replacement of IC18.

Figure 2:

A: PCB layout of the loop amplified board, matching the pins of the dual OpAmp BA718.

A capacitor (C1, 100 nF) was included across the voltage supply.

B: the finished unit before soldering into the PLL board.

C: PLL board after installation of the new board replacing IC18.

The evaluation of the LT1028 modification started with the original PLL board, without the Kenwood modification. The results are given in figure 1 (green line). At lower frequency offsets (5 to 20 kHz), the LT1028 modification appears to produce less phase noise while at higher offsets (>50 kHz) the Kenwood modification appeared to be slightly better. Both modifications in place (figure 1, red line) produce a limited further improvement above the single mods.

After these modifications, some weak spurs became detectable during the measurements which I not recognize before, because they were covered by phase noise. As to my knowledge, this is quite normal for conventional PLL designs. (By the way: the origin of one spur, which appears when a strong carrier is about 20 kHz off-passband , is interesting: this one results from coupling of the 20 kHz magnetic field from T1 on the DC-DC-unit over a long distance into the chokes L65, L69 and L72 on the RF board , causing a slight 20 kHz modulation of the VCOs.)

4. Comparison with other rigs

I was curious how the modified TS-940 compared with other rigs. Since I don’t own one of the contemporary top performance (and top expensive) rigs, I use an FT-857, which I recently bought.

It is technically similar to the FT-897. As figure 1 (yellow line) shows, the FT-857 is superior to the original TS-940, but the TS-940 modifications reduce its phase noise clearly below that of the 857 (exception: 3 kHz).

The phase noise levels of the OMNI VI+ an K2 transceivers for a 10 kHz spacing are -123 and -126 dBc, respectively [1, ]. Although comparisons with performance data given in the literature are limited (different measurement techniques etc.), a comparison with figure 1 suggests that the

TS-940, with the above modifications, is still very competitive.

---

In summary, the phase noise of the TS-9 0 can be substantially improve . The modification described in Service Bulletin #917 is highly effective. It can be even improved by replacing IC18 by an ultra low noise OpAmp, such as the LT1028.

73 e Thomas,

DF5KF

5. References

[1] Raczek T: The DX Prowess of HF receivers. QEX Sept/Oct, 1-5 (2002)

[2] The ARRL handbook for radio communications. ARRL headquarters, Newington,

84^th ed., 10.1 ff. (2007)

[3] website: http://rfdesign.com/vlf_to_uhf/time_an _frequency/radio_impact_ultralow_phase/

[4] web document: http://home.pacbell.net/johngreb/k2phasenoise.p f

6.7 Information on Ultra Low Noise OpAmps available in 2008

The LT1028

Described as

LT1028 - Ultra Low Noise Precision High Speed Op Amps

looks very low noise when compared to other OpAmps:

http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1154,C1009,C1026,P1234

Compare these very low voltage noise figures to other conventionl OpAmps

Voltage Noise

1.1 nV/rt Hz Max. at 1kHz
0.85 nV/rt Hz Typ. at 1kHz
1.0 nV/ rtHz Typ. at 10Hz
35nV_P_-P Typ., 0.1Hz to 10Hz

Gain-Bandwidth Product

LT1028: 50MHz Min.