Weak audio on harvey-wells tbs-50d

[W4TWA]

After three weeks of sorting through a dilapidated tbs-50'd harvey wells bandmaster, i finally lashed it together enough to join along in the gulf coast mullet society net on 3885 tonight.

I asked art for some fairly critical sig reports, such as audio and strength. Art commented my audio sounded weak, possibly 50% mod on my end. I had expected as much after my scoping the waveform.
I  rebuilt substantially the entire rig with new components where need be. Audio module has had all paper caps replaced.

So here are symptoms. Loaded into a dummy load and output rf monitored on my scope, the modulation power does appear about 50 per cent. There is a pot to adjust the audio in the pre amp section. Even with this pot at full output, same weak mod. Then I added an amplified d-104 to try and wake it up. No particular difference.

Pleas send a newbie some suggestions for tracking this problem down, I.e.,where to start looking, and at what method of narrowing this down
I feel I'm 75% to having this as it should be if I can get past this obstacle.
Thanks
Paul
W4TWA

[WQ9E]

Paul,

Double check that you have the correct resistors for your supply voltage on the patch panel on the back of the TBS-50D.

Were the cathode resistors OK in the audio section?  The 47K screen/plate resistors in this section?

How much input power are you running?  Double check the metering circuit because faults there could cause you to run much more input than intended.  Does the output power seem correct for the amount of input you are running?

[KA0HCP]

The D-104 amplified mic is designed to work with Low-Z microphones, <5,000 Ohms impedance, found in solid state transceivers from the 1960's on.

The D-104 crystal or ceramic element, however is a very Hi-Z element which works well with older tube transmitters.  Astatic recommends a load of >1 MegOhm, which is compatible with the original TBS-50d specification.

Through long experience hams have found the D-104 elements may produce better modulation and quality when working into as much as 2, 5 or even 10 MegaOhm load.

Suggestions:

-R101, at the mic jack, is shown as 1 MegOhm.  Confirm the actual resistance.  

-Bypass the amplifier section in your mic base, and feed the element directly to the plug (as originally designed).

-Try replacing R101 with higher resistances as noted above.

[W4TWA]

Thanks for the suggestions guys, I'll try these today. I'm running the HW power supply, about 500 volts output and 425ish on key down. About 100 ma of plate current according to the panel meter. My cw output is about 25 watts, about 20 watts on phone.
My resistors on the rear panel are correct. I will try running a lower plate voltage to see if it helps. Seems I read another person had to reduce power a bit to get 100%.
I will also change to un-amplified mike.
I replaced all paper caps and out of spec resistors, although I havnt replaced the electrolytics in the audio circuit. I will do that also and check the results again.
Thanks fellas

[N2DTS]

Looks like a very nice audio design, no driver transformer...

6L6's in ab1 do not make much power I would think, and would want a high plate and screen voltage.
I would up the voltage and load the final up lighter.


What voltage is the rig running at, and what kind of power do 6l6's make at that voltage?
R119 will be very important, it sets the voltage on the driver and other tubes...
But you may be going into AB2 or class B, a scope should show that when you turn the gain up...
 

[W3RSW]

Very interesting pre amp. Both 6au6's are used as triodes, can't remember but might even be the low mu way of triode connecting a pentode. Also one has no cathode bypass. Wonder why a single 12ax7 wasn't used?   At least the low mu phase inverter has twice the gain of a single triode.  I don't think that amp ever had excess gain or even gain enough to run full modulation.  Let's put it this way; a brand new crystal mike of the era with full output might make almost 100% mod. , but I wonder who had the raft of 6au6's at ten cents each back in the day.  

Still, something doesn't seem up to spec.  Be interested to see what finally is found. Assume tube changing has been tried.  One half of the phase inverter might be dead.

Also for a noticeable increase in gain, place a 50uf or so at 35 volts cap across the 6L6 cathode resistors.  Distortion won't increase too much.  Looks like a lot of cost savings went into this rig.  Maybe the original design was too hot and stuff pulled to tame it rather than correct feedback and distortion problems.

[W3GMS]

Hi Paul,

If you have an audio signal generator use that instead of the mic so you can buzz things out with a scope.  Go stage by stage looking for a clean undistorted amplitude.  Since you want mic level audio from the generator you might have to  pad the generator output down to deliver a few mv at the first mic amp stage.   

Also look at your resting current on the modulators and make sure its in spec.  You can do that with measuring the voltage across each cathode resistor and then doing the calculation if they don't monitor the plate current with a meter.   Just statically, without applying any audio you could look at one tube and then the other.  With an RC coupled scheme to the grids, The modulators should not pull any grid current so that's why the bias on the modulators are so important.   If its set wrong, then the peaks of the drive audio will flatten since it can't support the current demand.  Cathode bias is used and if the resistors are correct you may try different tubes or swap them in their locations and see if the scope waveform changes.  Also look at the grids of the modulator and you should see the same AC waveform on each grid and pay attention to the peaks.   

As others have said, make sure all the DC voltages in the audio section are correct.   

I have refrained from offering any circuit improvements since I am "assuming" that this transmitter stock will deliver close to 100% modulation.  Other on here that own one can confirm that. 

Either load the secondary of the mod transformer with a resistor which is the same as the plate impedance of the final or do the tests into a dummy load.  Be aware, if you do it with RF present you may get some RF into the ground lead of the scope probe and see some things that are not really present.  For the plate impedance value just divide the plate current into the plate voltage and then use that value resistor load.  Maybe something like a 20W resistor. 

73,
Joe -W3GMS   

[KA0HCP]

Thanks for the suggestions guys, I'll try these today. I'm running the HW power supply, about 500 volts output and 425ish on key down. About 100 ma of plate current according to the panel meter. My cw output is about 25 watts, about 20 watts on phone.
My resistors on the rear panel are correct. I will try running a lower plate voltage to see if it helps. Seems I read another person had to reduce power a bit to get 100%.
I will also change to un-amplified mike.
I replaced all paper caps and out of spec resistors, although I havnt replaced the electrolytics in the audio circuit. I will do that also and check the results again.
Thanks fellas

'Nuff said!

I had to build replacements.  One fit in the old case, the other did not.

Don't assume the bakelite capacitors in the rest of the radio are good until you measure them!   I have two TBS-50d's and HW used whatever parts they could get; no standardization, and probably not the highest quality.

Keep plugging.  

[AB2EZ]

The phase splitter circuit is "interesting". Perhaps it is a standard design from that era, but...

The output of the top half of the 12AU7 is divided by 11 [i.e. 10k ohms / (10k ohms +100k ohms)] and fed back into the input of the bottom half of the 12AU7. Therefore, the two phase splitter outputs will not be equal unless the input-to-output gain of the top half amplifier and the bottom half amplifier (taking into account all of the feedback paths) are both equal to 11.

The shared cathode resistor, combined with the rest of the circuitry, results in positive feedback. To see this in the schematic: separate the two cathodes, and place a separate 1k ohm resistor between each cathode and ground.  In this modified circuit, if the grid-to-ground voltage of the top half 12AU7 goes more positive (i.e., more plate current in the top half tube), then the grid-to-ground voltage of the bottom half 12AU7 goes less positive (due to the 1/11th feedback from the output of the top half tube to the grid of the bottom half tube)... which means that the cathode-to-ground voltage of the bottom half tube goes less positive. If you, now, reconnect the two cathodes, you see that the voltage gain around the (now closed) loop from the top grid to the top cathode is negative... which means that there is positive feedback (if the grid to ground voltage becomes more positive, the cathode-to-ground voltage becomes less positive).

Stu

[W3RSW]

Not only is it a standard circuit from that era, it is self balancing.
It and the single triode, split load phase inverter are described on page 261, 1963 ARRL handbook and in other year issues.  They are both self balancing.

 If you'd like I can copy the part that describes the two triode action. But short version is that the upper triode's voltage is split across both resistors at following grid in series. That reduced signal is then fed to the lower triode's grid which then amplifies it to be fed across the lower following stage tube's grid in series.  This 180 degree opposite signal opposes that coming from the upper voltages across the same resistor to ground at the common junction thus reducing the signal applied to the lower triode grid.  Well I've almost paraphrased the whole article but omitted their references to specified capacitor and resistor nomenclature and their values.  Suffice to say the common following stage resistor can be same as or one half following stage grid resistor.  Both triodes have a common cathode with shared resistor and audio bypass cap.

The gain of the two triode inverter is slightly less than twice the single tube triode amp at same operating voltages.

The gain of the single tube splitter is slightly less than two (1.9xx) even if using a high mu tube a/c it is highly degenerative.

[AB2EZ]

Rick

I checked out the phase splitter circuits in my copy of the ARRL 1965 handbook: figure 8-4 on page 236.

There are important differences between the ARRL design and the HW design.

1. The ARRL design includes are 10uF bypass capacitor across the shared cathode resistor of the dual triode (which is a 12AU7 in the case of the HW design) tube. This capacitor eliminates the positive audio feedback that I referred to in my earlier post... which would be very problematic in the HW design.

2. The ARRL design employs a different approach for feeding output from the top half tube to the grid of the bottom half tube. In the ARRL design, R113 is comparable in value to
R112 and R115 (not 1/10th the value)... and, in addition, in the ARRL design R112 and R115 are joined together... and connect to ground via R113. Therefore the ARRL design has a much higher amount of coupling from the output of the top half tube to the grid of the bottom half tube... and there is also negative feedback around the bottom half tube (from plate to grid due to the shared R113 in the ARRL design), which is absent in the HW design.

It is the negative feedback (and the associated large loop gain) around the bottom half tube (missing from the HW version) that ensures that the phase splitter outputs are approximately equal and opposite (provided R112 and R115 are approximately equal in value).

As shown, above, the HW design is more likely to oscillate than it is to be a self-balancing phase splitter. If it doesn't oscillate, it is not likely to be a self-balancing phase splitter (with equal and opposite outputs).

Stu

[w1vtp]

Rick

<snip>

As shown, above, the HW design is more likely to oscillate than it is to be a self-balancing phase splitter.

Stu

Interesting you should say that - my aunt, W1UET had a HW TB50D and it would always come on the air with a very strong squeal for about a second.  Then it would OK - might have been a electrolytic charging up.  I say OK, I don't recall hearing her on AM when she was in Rutland and I was in North Clarendon.   Back then, we pretty much accepted yellowy audio as the norm.  So, it wouldn't have been a stretch that she had low audio in addition to the squeal.

BTW, have you guys noticed that TBS50's are bringing a ridiculously high price and they hardly ever include the power supply.

Al 

[W3RSW]

Stu,
Yes, very interesting. There is a difference Where the ground and star jct. is off that 10k resistor. Now I wonder if the schematic and the actual circuit were supposed to follow the circuit as shown in the handbook and used in several radio AF amps of the period.  Perhaps the amp is ok but the schematic drawn wrong. Worth checking under the hood.

A rewire of the junction and addition of cathode bypass would be easy to install.
Low output does seem to indicate that half the output circuit is loafing.

[AB2EZ]

Rick

Agreed!

Add the 10uF bypass capacitor, change the 10k ohm resistor (R113) to 100k ohms,  and rewire so that R112 and R115 join together and then connect to ground via the new (100k ohm) R113.

The changes will ensure that the circuit works as a balanced phase splitter.

If the gain from the microphone to the inputs of the modulator tubes is not sufficient, then there are additional simple changes that could be made. For example, using a 12AT7 as a pin compatible substitute for the 12AU7, and with a 270 ohm cathode resistor (R110) to get about 10dB more gain.


Stu

[W4TWA]

Wow, fantastic. I'm overwhelmed at all the responses, and enjoy being taken to school by you virtual Elmer's, haha.
I have taken it down to the bench and replaced the operating position with a ranger 2, while I work on the Harvey.
This radio has been a labor of love. I have spent weeks tracing and un doing all the mods. It was modified to not work apparently. Botched ptt switch, and several other patently obvious rewires that rendered it dead as a door nail. I was thrilled to have it on the air since god knows how long, but will not rest till I'm getting the audio it needs. Back to the shop tomorrow and sort through these suggestions, starting with the easiest first.
Thanks to all

[KA2DZT]

I agree with Stu,  first thing I noticed was the strange voltage divide with the two resistors in the phase inverter.  I didn't comment on it yesterday to not confuse the situation.  The whole audio section looks a bit strange.

I have to go back and take another look at the schematic.

I took another look, I think the 100K/10K voltage divider is correct given that 1/10 the voltage output is feed back the grid of the second half of the 12AU7 to be amplified.  A voltage gain of 10 is about right for a 12AU7.

Adding some cathode by-pass caps on the cathode resistors should bring up the voltage gain of the whole audio section.

A single cathode resistor for the two 6L6s is standard.  I think there are two resistors but they are in parallel, but they should be by-passed with a cap (25ufd 50V or anything close).

Took another look,  the shared cathode resistor for the 12AU7 is another thing that doesn't look right.  I agree with Stu, I think there should be two separate cathode resistors.

Fred

[AB2EZ]

Fred

et. al.

I was not suggesting that the shared cathode resistor of the 12AU7 should be split into two separate cathode resistors... only that it should be bypassed by a 10uF capacitor. However, splitting the currently shared cathode resistor into two separate cathode resistors (each twice the value) would be a good way to further ensure that there is no positive audio feedback around the circuit. Keeping the shared cathode resistor... if properly bypassed... is also okay. At this point, since the cost of the parts is not an issue, I would go with your suggestion of using two separate cathode resistors (each twice the value of the current shared cathode resistor)... and I would also use two separate 10uF cathode resistor bypass capacitors.

If the feedback circuit (R113, R112, and R115) is rewired to match the ARRL design...as suggested... then the 10k ohm resistor should be changed to 100k ohms. This is because the resulting negative feedback around the lower half tube (missing in the HW circuit, but present in the ARRL circuit) needs to be high enough to cause the circuit to be self-balancing. With a 1/11 voltage divider (in the existing circuit with R113 set to 10k ohms), and the modest amount of voltage gain in the 12AU7 (somewhere around 11), there is insufficient gain around the lower half tube's plate-to-grid negative feedback loop for the self-balancing to work properly.

Even with R113 changed to 100k ohms... the circuit will not work as well as the ARRL design is intended to work... because the open loop gain of a 12AU7 is too low (see my additional, more detailed comments in my next post).

Stu

[AB2EZ]

For the benefit of those who are interested in the details of how the ARRL circuit works:

The top half triode provides a voltage gain from its input (grid-to-cathode) to its output (plate-to-ground) of: A (volts per volt). For a 12AU7 (with the cathode resistor bypassed): A is somewhere around 11.

The voltage divider directs a portion of the top half tube's output to the grid of the bottom half tube. This portion is B. The value of B is 0.5 in the ARRL design. The value of B is 1/11 in the HW design.

In the ARRL design, but not in the HW design, there is also (negative) feedback from the output (plate-to-ground ) of the bottom half tube to the input (grid-to-cathode) of the bottom half tube. The gain around this feedback loop (in the ARRL design) is A x B.

Therefore the gain from the output of the top half tube to the output of the bottom half tube is:

[Vout bottom] / [Vout top] = the voltage divider ratio x [the closed loop gain of the bottom half tube] =   B x [-A /[1+ (A x B)]

If A x B is large compared to 1, then the output of the bottom half tube is -1 x the output of the top half tube (i.e. the phase splitter is balanced).

In the present case, if we rewire the circuit to match the ARRL design, and if we change R113 to 100,000 ohms... so that B is around .34 (taking into account the relatively low plate resistance of the 12AU7), and if we assume A (for a 12AU7) is around 11, then A x B is around 3.7. This is not as large as would be desirable, but still okay to obtain a marginally balanced phase splitter.

If one substituted a 12AT7 for the 12AU7, and made each of the two (separate) cathode resistors 470 ohms (to get the proper biasing for the 12AT7)... then A would be around 60.

This would produce an improvement for two reasons. First, the top half tube would have around 15dB higher gain. Second, the gain around the feedback loop of the lower half tube would be much higher... leading to much better balance in the phase splitter.

Returning to the HW circuit:

Since there is no feedback around the lower half tube, the circuit is not self balancing. If the open loop gain of the bottom half tube is approximately 11... then the output of the lower half tube will be approximately -[1/11] x 11 x the output of the upper half tube= -1 x the output of the upper half tube. Since the 12AU7 has such a small open loop gain, and such a low plate resistance... if one sticks to the 12AU7 (rather than switching to a 12AT7), one is probably better off using the HW design (except for the addition of the 10uF cathode resistor bypass capacitor)

Stu

[KC4VWU]

A problem I found with one here is flakey tube sockets in the mic amp sub chassis. There was one that just wouldn't behave no matter how much I tried to clean or re-tension; it would still give low audio output. I finally just replaced it and problem fixed. It was probably a bad ground connection due to the sockets being riveted in. Happy Holidays!

...Phil

[w4bfs]

interesting discussion .... I tend to prefer the long tailed phase inverter which works fb ... and RCA included a triode connection for the 6au6 in 92cm-6854r1