1969 Knight R-195 Communications Receiver Kit

[Smoky]

TR-1, TR-2, TR-3:



The back of the transistors have paint markings so they were graded or matched to some degree:



I watched a Youtube video as to how a JFET can be tested using a DMM.

Using a Fluke DMM, here are the three transistor's resistance results between:

TR-1

-source and drain is 130 Ohms
-gate and drain is 1.48 KOhm
-gate and source is 1.48 KOhm

TR-2

-source and drain is "open"
-gate and drain is 1.28 KOhm
-gate and source is "open"

TR-3

-source and drain is 160 Ohms
-gate and drain is "open"
-gate and source is "open"

Something isn't right with one or two of those transistors. Btw, someone gave me their Triplett analog meter but I don't have leads for it. Why? because the leads are required to have "female" plugs on their ends! Yes, those are "male" banana pins on the inputs of the meter



Anywho, I found two sellers on Ebay that are selling new National and Motorola MPF-102 FETs. I'm going to buy a dozen or so.

Btw, I believe I can notch the brass disc on the S-1C wafer on its left side to give it enough clearance without taking the entire Band switch apart

[floobydust]

I would do surgery on the rotary switch contacts, they're supposed to point straight to the center and a few look bent. The camel hump needs to be there too.
Sometimes the rotor nails the side and mashes and bends the contacts. Using a magnifier or microscope, I just straighten the top and bottom leafs, and have them leaving a small air gap the rotor can swing into.
I usually clean them dragging a small piece of printer paper through. In the TV repair shop, we'd also lube the rotor with Lubriplate but Fluke 87I uses the same style rotary switches and they recommend Nye Lubricants 813S which is a fluorocarbon gel. I don't think they work completely dry.

What the JFETs just died, performance anxiety or all that awful music on AM shocked them lol
While measuring ohms D-S, the gate cannot be floating or you will get mysterious readings.

[Smoky]

Got it FloobyDust.

As the Band switch is turned, that notch in the brass disc splits all of the contacts down the row so it's not just at pin 1. The contacts look to be aiming to the center in real life. I'd hate to loosen that rivet that pinches those contacts together then, if they're hit sideways, they'll spin for sure.

I'll gap all of  the contacts like you said so that the brass disc has an easy start slipping under them. I looked with a microscope at them already and I can see green tarnish under some of them too!

As for the FET resistance tests using my DMM, the results settled down solid. I just checked them again.

Why those FETs failed is beyond me. I do recall hearing snaps and crackles through the speaker when the Band switch is turned. Too bad there isn't a way to shut down the power to the switch until the selected band rests into its pocket/detent. It's probably slightly arcing while the Band switch is slowly turned across the contacts and maybe that's enough to shock the FETs?

Anywho, we'll get this radio right

The 15 new National MPF102 transistors came in:



What looks promising is that the back of the National transistors have paint markings too:



I moved over to the Band switch to notch the brass disc on one of the wafers. The disc wasn't indexed correctly by the factory so I needed to clearance the left edge of the notch to keep the disc from shorting two terminals at the same time. What a job! I used a Lenox Gold utility blade to make over 200 slices across the brass disc. I first clamped the disc so that the notch rested in-between two terminals. I then slipped a thin piece of aluminum under the disc to stop the utility blade from cutting the wafer material as it crossed-over the edge after each pull of the knife



Here you can see the small amount of brass starting to lift:



I removed the small burrs and polished the disc with 1000 grit sandpaper. This picture shows the Band switch at position 1. The terminal is now isolated from the other four, and that also goes for the rest of the terminals down the line when the Band switch is turned



My new DCA75 arrived today too! Time to measure-up some JFETs and see if we can't resuscitate this R-195



...and while the radio is down, I decided to refinish the pointer. When I tried to adjust the pointer to the chassis, the paint would just crack or peel off entirely. Sure enough, I dunked it into some methyl chloride stripper and found out the base metal is brass. I'm going to scuff it and mist it with white epoxy primer and top it with a gloss white urethane

[floobydust]

That's some switch surgery alright. I would have a nervous breakdown doing that   Does it pick up WWV?

The JFETs have a 10:1 IDSS variance 2-20mA as well as transconductance 2,000-7,500 umhos.
It makes a kit hell unless the JFETs were hand selected in the first place. You had one bogey (three stripes) different than the other two (two stripes).
RF amp TR-1 looks around 4.3mA, mixer TR-2 with 2.0V/10k is only 0.2mA, and TR-3 local osc about 5.6mA. So mixer TR-2 seems to be running at very low drain current. The DC voltage reading at TR-2 might be misleading due to RF present from the local osc. TR-3. I guess you could jam the LO by shorting R-7 or adding a big cap and see if the reading on the mixer changes.

Knight might have had issues with the local osc. overloading the mixer? I note in the original post handscribbled schematic from Bill Meacham, the guy found R-4 was 50k and only 1.0V there, which is really small drain current, as well as R-7 as 47k. It is a puzzle.

[Smoky]

Thanks for your help FloobyDust!

Here are the voltages of the two original Motorola MPF102's and the new National in each band. And just like you said, RF noise may be present. Also note, Knight interchanged the Source and Drain connections of the JFETs in this radio too:



For an experiment, I matched three new National JFETs with the DCA75.  They have nearly the same Vgs "on" and "off" voltages and so I installed them to see how they affected the performance of the radio.



Well, the performance got worse. The local radio station still comes in but it's edgy and gritty and the station was harder to lock onto. Here are the new voltage measurements below. Are the new JFETs pushing the radio into distortion/clipping?



The voltages moved outward to the higher and lower extremes but stayed in proportion. There is a pattern here, unfortunately, there are no notes anywhere on the schematic as to what the conditions of the radio should be in when taking voltage readings.

I'm not sure if this drawing applies to how the JFETs in this radio operate, but it depicts the Drain voltage resting mid-point between the Source voltage and the supply voltage while at idle:



Once the JFETs are brought to their idle state, couldn't the surrounding resistors be tweaked to bring the JFET to its correct bias?

It reads that the lower Source voltage should be no less than 2v at idle, and knowing that the supply voltage in this radio is 8.70v, the Drain voltage should be set somewhere around 5.35v.

Or am I'm on the wrong track and maybe just swapping JFETs would be a better approach?

Who knows? and what are the odds that the printed schematic voltages were taken while a signal was applied to the antenna terminals? wouldn't that bring the upper and lower voltages of the JFETs to their near-maximum limits before distorting

*We have plenty of options and my goal is to get the radio working the best it can. The radio works. The new resistors are surely more accurate than the old ones and the old resistors wandered in value. Maybe, the two original/working JFETs should go back in and we just deal with "grading" another for TR-2's slot? The copper traces are super-tough so component swaps are not a problem. In another article, someone made a MPF102 tester using a breadboard and simulated the circuit it needed to work in.

And about Bill Meacham's findings, He described the (Drain resistor?) R4 change as "slightly noisier."  Didn't that resistor adjustment also push the Source voltage even higher (I heard that resistance changes make the D&S voltages raise or lower in unison) and closer to the supply voltage causing the signal to go into clipping? Shouldn't have R6 been adjusted too to counteract that?

Up to this point, the only thing we have touched is the JFETs. I have not turned any coils or touched/replaced any other parts except for the new resistors.

...and I'm in no rush in getting this radio done



The more I look at the numbers of the matched JFETs, they don't look so bad compared to the voltages on the schematic, except for TR-1's voltages under the "Source" column. Hmm. We got Bill Meacham's issue

*One other thing, the Band switch is on "A" in the schematic so look only across that line when comparing the JFETs voltages.

[floobydust]

JFET's wide variance on three parameters is something that really confuses people, makes it messy and people lose patience.
Matching JFET's is usually only necessary in say an audio diff amp pair, and threshold or cutoff voltage VGS(off) is just an offset, matching that is not the gain or IDSS so I wouldn't expect it to help here. You might have just ended up swapping in parts that have lower gain.

What I'll do is just breadboard a simple common-source amp with the same values, inject 1kHz sine-wave and measure input/output voltages to rank the parts for gain and drain current etc. This ends the madness of guessing what's going on. Does the DCA75 give those numbers? It could. Should fire off an email to them.

I think what you want is the highest (gain) JFET transconductance (input voltage controls output current) YET with IDSS and VGS(off) that won't cause the stage to bias off at one end and clip/saturate/distort with the circuit's component values.  But across the board higher gain, TR-3 greater LO amplitude- could overload the mixer. I don't know what a design goal for that is, for radio. Might be a balance with R-7 and R-4 setting LO amplitude to the mixer, but R-4 also affects many things. I thought you'd want R-4 low as possible for high mixer gain but that scuttles running the stage at very low current.

Maybe consider using transistor sockets for now. Another approach is too look at schematics of later Knight kits and see if they made changes. Or simply try change a resistor value.
W. Marshall Leach Georgia Tech course: ECE3050 The JFET has math for those interested.

[bd139]

Quick note on the MPF102's. They are garbage FETs generally. Wildly varying in Idss and half of them didn't even work to start with. Their popularity was down to availability at the time and cost. This and usually being in stock by Radio Shack caused a somewhat cult following in ham radio circles and the inability for people to understand the circuit and to substitute parts raised the associated cost.

It's best to look at the bias conditions (just chuck the circuit in LTspice) and grab a better JFET for the job really. The "graded" parts from the same process are actually now known as 2N5484/2N5485/2N5486 which are far more reasonably priced and avoid the whole selection and measurement processes the MPF102's tend to require.

[Smoky]

This is embarrassing

FloobyDust, I did absolutely that, I swapped in lower gain parts!

So I put one of the original Motorola MPF102 JFETs on the DCA75:



IDSS=10.51mA
Rds=148.5 Ohms

Next, I removed one of the three matched National JFETs from the radio:



IDSS=7.25mA
Rds=137.4 Ohms

Now get this. I went back to the remaining 12 National JFET's and I tested them again, but this time, I looked for the highest IDSS:



IDSS=12.00mA
Rds=106.2 Ohms

I found two at 12.00mA and one at 11.7mA. What's interesting too is that their resistance readings are between 106.2 and 109.1 Ohms.

These babies will be installed in about an hour

Wait a minute! Look how the symbol changed after IDSS on the DCA75 analyzer. It went from equal (=) to greater than (>). This means the gain is even more than 12! Hell yeah

[KE5FX]

If you haven't hooked your DCA75 up to a PC and played with the curve tracer feature yet, I'd recommend it.  It's very nifty, and very educational.

[floobydust]

WAIT another minute! IDSS is not the gain - Look for highest "gfs" as Peak calls it.

Quick note on the MPF102's. They are garbage FETs generally. Wildly varying in Idss and half of them didn't even work to start with. [...]

Of that late 1960's era, MPF102, BF245, 2N3819, 2N5457, 2N4416 etc. were out there and I can't see anything that says they are poor semi's.
JFET's look almost all the same in datasheets. Huge variance and nobody stomps the competition. Only Toshiba refined them in the 80's.
I think any low performance might be due to simple age of the part, NOS isn't always new or the best of the lot.

[Smoky]

Thanks for the help FloobyDust.

I put the two original Motorola JFETs back in their original locations (TR-1 and TR-3) and a new National in TR-2's spot and that's all the radio needed to bring it back to life

When you said that the transistors may have been hand-selected, I wasn't going to take any chances.

All five bands pick up stations even though I'm using just an 8' wire as an antenna!

[floobydust]

You mean all these years I've been tuning in WWV Colorado 10MHz 1,100 miles away instead of "CHU"? It sounds bland in comparison and it ends up farther away. Atmospherics wipe out the signal some days and I'm just using a jumper lead for an antenna.
With the S-meter near pegged I think it's working well 
Are you going to tweak the IF or anything. I have no lab room for a service receiver... although they would make alignment easy, and eBay has them low cost... and winter is coming around soon...

[Smoky]

It's amazing how many stations there are! Tune the dial and then the Bandspread and Antenna knob and they come out of the woodwork. Every little nick and tick is a station.

Tons from South America it seems, or Cuba, or both!

Anywho, I was tuned into CHU Canada at 7850KHz. Ottawa is about 900 miles due north of Raleigh, North Carolina.

I recall when I had the front dial scale on the receiver a week or so ago, our local AM station 680 was playing almost an inch off from the numbers. I image there's plenty of tuning to do. I do have that HP spectrum analyzer now so I'll soon be looking for some advice


...and shifting to another topic, I went back and measured the 9v rail in the radio at CR-10. It measured low because the initial batch of JAN1N757A-1 Zener diodes I purchased measured (~8.65v). The 1N757A's are 5% tolerance parts and aren't available any tighter unless they're "special" ordered, so it's a crap-shoot:



I wanted to see if I could get the 9v rail closer in line to the schematic or slightly higher so I purchased another batch of Zeners. This time I bought JANTX1N757A-1's:



Sure enough, the voltage climbed a little higher!



After the diode was installed, I measured the 9v rail again:





I hear a noticeable difference in the reception/clarity. One band doesn't sound as saturated as before and I can see that the needle pegs the S-meter more sharply. There is a meter adjustment pot on the back of the radio if I need to tone the "swing" down.

Anywho, I'm getting to really like this Peak DCA75 analyzer, it has saved me a ton of time already

Plus, the pointer is re-finished and looking goooood!



I added new rubber feet to the case too. The front pair of feet are slightly taller than the rear to improve the view of the front panel. I bought them from www.RubberFeet.us and they are very nice quality and made in the USA!



As soon as the dial and faceplate are back on the radio, it'll be time for an alignment. I just don't have the proper RF signal generator yet.

I didn't know that the Siglent SDG1032X is also an RF signal generator too! So when I'm back from the Shelby, NC Hamfest this weekend to find some old radios to fix, it'll be alignment time

[floobydust]

You might lower the RF gain or check the AGC action if it's distorting on strong signals, or maybe the S-meter cal is a bit off.
Strange the schematic lists RF Gain voltage (39) at 0V but RF gain pot set to -4V and supposedly +8V other side of R-45. The wiring to meter driver TR-9 makes no sense, there must be errors in the schematic, something has to connect to AGC (55) like R-18. R-35 is a C-35? lol, and it should have reverse polarity + to GND (you prob. used a 1uF film cap there anyhow). AGC seems to be node (55).

Back in the day I can imagine communicating with draftspeople in Japan being a major difficulty, to fix errors in the schematics or documents.

P.S. Appreciate the high quality pics you post.

[Smoky]

Knowing now that the Siglent SDG1032X does AM modulation, I realized that the waveform generator only goes to 30MHz. When modulating at 1KHz, wouldn't the Siglent need to go beyond 30MHZ? 30MHz + 1KHz.

I haven't started any part of the alignment procedure yet but one of the steps requires the signal generator to be set to 30MHz modulated 30% at 400cps or 1000cps (the highest setting required in the alignment procedure for the Knight R-195).

So I set the Siglent to 30MHz, modulated 30% at 1KHz:



Having picked up a used HP3588A Spectrum Analyzer recently, I wanted to see if I could capture the signal and have it tell me if the Siglent will work. I set the analyzer up the best I could to get a clear shot. I put the marker on each peak of the modulated signal. Here's a shot with the marker at the top of the center peak (30MHz):



Next was the marker over the left peak (30MHz - 1KHz):



And now the marker over the right peak (30MHz + 1KHz)



All of this is just a learning experience and probably not absolutely required but it's good to get a "visual" of what's going on and to see that the Siglent will go a tad over 30MHz accurately and evenly.

[KE5FX]

Note that it doesn't have to tune past 30 MHz to generate an upper sideband above 30 MHz.  You'll get the sidebands for free when you apply either AM or FM modulation (or any other kind.)  That's literally the idea behind modulation.

[bd139]

WAIT another minute! IDSS is not the gain - Look for highest "gfs" as Peak calls it.

Quick note on the MPF102's. They are garbage FETs generally. Wildly varying in Idss and half of them didn't even work to start with. [...]

Of that late 1960's era, MPF102, BF245, 2N3819, 2N5457, 2N4416 etc. were out there and I can't see anything that says they are poor semi's.
JFET's look almost all the same in datasheets. Huge variance and nobody stomps the competition. Only Toshiba refined them in the 80's.
I think any low performance might be due to simple age of the part, NOS isn't always new or the best of the lot.

Not that at all. MPF102 wasn’t binned basically. The 5456/57/58 were the same part but binned. My point was pick the right binned part rather than the dubiously specified MPF102 so you don’t have to select it yourself. Sometimes when you throw an MPF102 in circuit the bias will be shot due to the Idss variation.

3819/4416 were same part but 4416 was selected and packaged with grounded TO18 case.

Age isn’t really an issue with these I’ve found. I’ve got a 1969 dated bag of 2n4416’s here and they’re all fine.

Regarding “modern” MPF102’s, the past is throughly discontinued so you have no idea what you’re getting. The binned variants are still available though I understand but because they don’t have the magic MPF102 written on them and people don’t know how to select and substitute parts they command less value.