Help with Carvin Legacy problem?

[nickdanger00]

Greetings...
I scored a used Carvin Legacy last month for a very good price. Now, either because it was defective when I bought it or it suffered some damage due to some tube substitution errors that I made, I am having trouble with it....

The symptom is that the gain channel doesn't. It used to have several tons of gain on tap and now it has a couple of ounces. The clean channel seems to be fine. I had tried substituting some lower gain tubes in the V1 and V2 positions to see how it sounded. These substitutions included two 12AU7s. This was a mistake, as after running the numbers on the circuit, 12AU7s could overdissipate the plate resistors on the first four stages.

I open the amp and expected to find some or all of these out of spec. They are not. However, there were issues. First, 4 out of the 5 preamp tubes read "Replace" on my little emissions tube tester. One of the four output tubes also read in the "Replace" range. After getting good tubes in, I ran a DC voltage check. Rail and plate voltages for all but one of the gain stages were nominal with respect to the schematic (http://www.carvinmuseum.com/pdf/amps...gacy%20Amp.pdf). The plate voltage for V2B measured about 100 Volts too high. This is despite the fact that the plate and cathode resistors measured nominally.

I ran a 1KHz sine wave into the amp and the first gain stage (V1B) managed a gain in the 40's, which I would expect. The second gain stage's (V1A) gain was 0.3 and the third gain stage's (V2B) gain was about 4. Clearly these are not what one would expect.

Thinking that I might be looking at one or more leaking coupling caps, I measured the DC voltage at the grids at idle. While not zero, they were less than 10mV. I measured these with the tubes in the sockets. After a little research I think should have does the measurements without the tube in place which I will do.

I also did a continiuty check of the tube sockets to the PCB and these seemd fine. The solder connections for the sockets all look pretty solid.

Since this is a PCB amp, I did not want to start replacing parts willy-nilly as there is always the risk of wrecking it but good.

I also saw this is as a good learning experience, so is there a sensei out there who can offer some advice about what I could check next?

Thanks.

[hasserl]

Check your filament voltages. My experience with Carvins has been that often the preamp filaments get less than 5 volts. Many 12AX7's will not light up and work at those low voltages.

That may not be your problem, but I would sure check it before you spent a lot of time checking other things that may not be a problem.

[nickdanger00]

Hi and thanks for that.
I did check the the filament voltages and they look high enough (V1-V3 @ 5.47VDC, V4, V5 @ 5.59VAC, V6-V9 @ 5.16VAC).

With the HV on with no tubes inserted I found that there was a fluctuating voltage at the grid of V2A of +/- 1mV. The rest of the grids read hard zero volts except the grid of V3A which read a steady 0.5mV.

I am now suspicous that C26 and/or C27 may be leaky. These should be pretty easy to change out.

If you have another opinion, please weigh in. In any event, thanks for the help.

[Enzo]

A few millivolts is not a problem

Trace signal through it. APply a steady signal to the input and follow it though the amp channel. USe a scope. No scope? Use an AC voltmeter. Or make a signal tracer with another amp and a cap at the input to it.

[nickdanger00]

Thanks for the info, but....

A few millivolts is not a problem

Trace signal through it. Apply a steady signal to the input and follow it though the amp channel. USe a scope. No scope? Use an AC voltmeter. Or make a signal tracer with another amp and a cap at the input to it.

Thanks for the response!

I did run a signal through it as you suggested. My orginal post describes the results. I ran a 1kHz sine wave into the amp, adjusted the level so the voltage at the grid of the first gain stage was 20mV, and dimed the drive control but zeroed the volume. I measured the voltage at the stage's grid and the voltage at the output side of the coupling cap to the next stage. I divided the output voltage by the input voltage to compute the gain of each stage.

The first gain stage passed signal with a gain of about 40. The second stage passed signal with a gain of about 0.3. The third stage passed signal with a gain of about 4. The fourth stage passed signal with a gain in the high 30's. The signal is making the rest of the way through the amp since I can hear it if I move the volume up a smidge. Stages two and three are clearly under-achieving.

While I suspected that DC grid voltages fractions of a mV from zero are not a cause for concern, it is good to get some confirmation on this. However, the voltage at the grid of stage three is not steady. My DC meter shows it fluctuating between +/- 1mV. When I put the meter in AC mode it reads a fairly steady 19 mV. This may not be my whole problem, but something upstream is not behaving itself.

Thanks again for your response!

[Enzo]

Nah. Again, it is a couple millivolts - a tiny bit of noise.

Gain of 0.3??? FInd out what is wrong there. You measured signal voltage at the grid and at the outboard end of the coupling cap where there was none. Measure it at the plate. A bad coupling cap would not be cosmic.

[nickdanger00]

Good advice....

Nah. Again, it is a couple millivolts - a tiny bit of noise.

Gain of 0.3??? FInd out what is wrong there. You measured signal voltage at the grid and at the outboard end of the coupling cap where there was none. Measure it at the plate. A bad coupling cap would not be cosmic.

Thanks Enzo, your advice is good. I'm embarrassed that I didn't think to do this myself. Let you know what I find.

[Enzo]

Use your third eye.

[nickdanger00]

Nice one....

Use your third eye.

Oooh... Very good. That one almost slipped past me.

Sadly, my third eye is blind.

Actually I do have a question... When I said that I was going to take your advice, I thought I knew how to do it. However, now I don't think I know.

When you measure the signal on the output side of the coupling cap, it is ground referenced, so the other probe is connected to ground. When you measure the output before the coupling cap, the voltage is referenced to ????. I.e. where do you put the other probe? When it is grounded I read very small voltages, even on stages that appear to be working. If I measure across the plate resistor, which was my next guess, I also read very small voltages. When I measure across the tube, i.e. plate to cathode I get a non-trivial voltage reading, but I'm not sure what it means.

Can you tell me where to put my probe?

Sorry, I couldn't resist that one.

[Enzo]

Your signal is an AC voltage.

Stop. I don't want to get into the argument that it is not AC, but fluctuating DC. We all know what it means.

Set your meter to AC. Now if there is 2 volts of signal, it will be 2 volts whether it is at ground potential or if it is on a plate at 200vDC. On AC volts, your meter blocks DC. In the circuit, the point of a coupling cap is to couple the signal stage to stage while blocking the different DC levels.

Reference the meter to ground. Black probe to chassis, red probe to plate or whatever. The signal is there at the plate or not. If it is there but not on the other end of the cap to the next grid, then obviously the cap is bad. If there is no signal onthe plate, then the tube is not functioning.

Also if that tube is a cathode follower, it won't have gain. Nor signal on the plate.

[nickdanger00]

Cockpit error....

Your signal is an AC voltage.

Stop. I don't want to get into the argument that it is not AC, but fluctuating DC. We all know what it means.

Set your meter to AC. Now if there is 2 volts of signal, it will be 2 volts whether it is at ground potential or if it is on a plate at 200vDC. On AC volts, your meter blocks DC. In the circuit, the point of a coupling cap is to couple the signal stage to stage while blocking the different DC levels.

Reference the meter to ground. Black probe to chassis, red probe to plate or whatever. The signal is there at the plate or not. If it is there but not on the other end of the cap to the next grid, then obviously the cap is bad. If there is no signal onthe plate, then the tube is not functioning.

Also if that tube is a cathode follower, it won't have gain. Nor signal on the plate.

Thanks Enzo, your reply confirms what I thought should be happening and how I thought I should set my meter. You're not going to get any philosophical arguments from me.

I now think that my whole AC signal measurement test from before is tainted. I was using a 1KHz input signal, but I looked at the specs of my brand new meter and it only measures AC up to 400Hz. There is also a discrepancy between the mV and V ranges on the meter that I have to sort out (perhaps by chucking the meter and using a better one). In any event I will have to start over.

In an attempt to shotgun the problem I replaced the plate resistors and plate resistor bypass caps for the first four stages (no cathode followers) as well a couple of coupling caps. All of this was to no effect. I don't think it is the tube that is the issue since the higher than specced plate voltages are identical no matter what tube I put in. I now wonder whether some of the PCB mounted tube sockets have been damaged with all of the tube insert/remove operations that I have done.

I will re-run the AC measurements using my better understanding of my meter's behavior or using a different one. I will also check the continuity of tube sockets to PCB. I will also re-check the quality of the tubes I'm using and look for shorts.

Any other ideas?

Thanks again for your help.

[Enzo]

This one is simple. Ground the meter, set on ACV, probe one end of that cap, then the other. Though teh DC voltage is very different on the two ends, the signal should be the same.

Your meter may read only to 400Hz, but it probably still will measure 1k, even if it rolls off some. Let us say the signal at 1kHz is 10v, but the meter only reads it as 8.2v. We may not know what the true voltage is, but the meter will still measure that 8.2v whenever it sees 10v of real signal. SO that would be good enough to see if the same voltage is at both ends of a cap. For work like we are doing, we are looking more for where the signal gets lost than what its true level is. It is here, but not here sort of thing.

Always consider what the tool is used for. My bench scope is an old 35MHz B&K - nothing fancy. I sometimes work on wireless mics. Those transmit on about 170MHz. If I clip my scope probe to the transmitter wire, or sometimes just wrap a loop of wire around the transmitter a few times, when I turn the thing on, I get the teeniest little deflection on the screen. Not enough to resolve a waveform. but consistantly enough to see. And that is all I need to verify the transmitter is sending SOMETHING out.

SO my 35MHz scope can dimly see 170MHz. I bet your 400Hz meter will still see some 1kHz. It just won't be accurate.

[nickdanger00]

OK, are you ready for this?

Well I found the problem... There is an intermittent connection in one of the tube sockets. I've found that if I seat V1 in the socket completely, the problem occurs. However, if I "tip" the tube over a bit in the socket,... the problem goes away. It seems that one or more of the second stage's pins are not making a good connection when the tube is fully seated. This is why I was seening extremely low gain in that stage.

If there is an intermittent connection on the V2 socket as well, then this may also explain the higher than expected voltage I am seeing at the plates of V2.

Either this has been a problem since before I got the amp, or I caused it when I was trying some tube swaps.

'Course the the question now is what to do about it. I guess I can buy a couple of new PCB mount sockets and have a go at replacing them.

Is there another alternative?

Thanks.

[Enzo]

Squirt cleaner down the pin holes and work a tube up and down in it.

Tighten the pins

Resolder the socket

Replace the socket, they only have 9 pins.

[nickdanger00]

Makin' it tight....

Squirt cleaner down the pin holes and work a tube up and down in it.

Tighten the pins

Resolder the socket

Replace the socket, they only have 9 pins.

Thanks Enzo...

Replacing the socket might be the best solution unless there really is a way to tighten the little gizmos that grip the pins. The one for pin two is clearly spread open much more than the others and is pretty much the cause of my problems.