Seventh Circle N72: odd distortion on scope

[soapfoot]

Just built a few of these, and during calibration, noticed that one of them was distorting oddly.

This is with a 0.5Vpp 1kHz sine input. Shot of my scope attached (I didn't listen to it, but something is clearly not right). All transistors Q1-Q6 appear to measure logical voltages. I went through and reflowed a bunch of joints and just really pored over it with a magnifying glass... checking capacitor polarity, parts orientation, etc... to no avail.

The one thing to note with this unit is that I had to do a bit of rework to the input transformer on assembly. I wonder whether I may have damaged the transformer somehow, and whether this type of distortion could be consistent with something like that?

Schematic here: https://www.seventhcircleaudio.com/_files/ugd/7ed8a6_d5b1b30ea44946bebbb02e1440baba94.pdf

This is the output:

 

[RuudNL]

How is the frequency response? Usually a defective transformer would influence the frequency response.
Next step is to check if the problem occurs before or after the potentiometer.

 

[Ike Zimbel]

Check for an open circuit on the switch contacts...it's also worth checking to see if you get this on every switch setting, so you will need to adjust your source signal as you check different gain settings. If it, works below, or above a certain gain/att setting, it points you towards where to look for the issue.

 

[JohnRoberts]

Does it clean up with lower input signal...

JR

 

[soapfoot]

Does it clean up with lower input signal...

JR

I just checked, and yes, in fact it does!

I'm not sure what that means yet, but it's definitely something...

Thanks to all so far for your help and ideas. I'll have time to dig deeper a bit later.

 

[Ike Zimbel]

It probably means you have too much gain. I would triple check parts values. I have trouble-shot a few kit builds for people and the "resistor value off by a factor of 10, 100, 1000 x etc." is pretty common.

 

[Ricardus]

It probably means you have too much gain. I would triple check parts values. I have trouble-shot a few kit builds for people and the "resistor value off by a factor of 10, 100, 1000 x etc." is pretty common.

How about off by a million?

I bought a kit once and it was supposed to have two 10 ohm resistors, and the dude shipped two 10 meg resistors!

 

[JohnRoberts]

Look for bad solder connection in loop gain setting network... wrong value could do it too...

JR

 

[swpaskett]

Remember you are testing a mic pre, designed for mV inputs with lots of gain. Hitting it with 0.5V will overdrive it if the gain is set much higher than about 30dB, giving you serious distortion. I think maybe this one is working right and the others may have a problem with low gain (or you had the gain set lower, which might be more likely). Work the math. If you set it for 40dB of gain, that's X100, so your half volt should be 50V on the output, probably a bit more than this board can manage unless they are counting on lots of transformer gain on the output. Set your input closer to 10mV and retest all the boards.

 

[soapfoot]

I’m following the calibration instructions in the manufacturer’s documentation

 

[swpaskett]

Okay, then there must be a fair amount of transformer gain on the output.

The very first thing to do is check the power supply. Make sure it is near 24V. If not, fix that problem, then move on.

If I am guessing right, you are on step 49 of the procedure, Fine Bias Adjustment. The desire here is to set the DC operating point at the middle of the power supply voltage. The power is regulated to 24V, so midrange is 12V.

The way they suggest doing this is to "gently" overdrive the input and adjust the bias so that the output exhibits the same level of clipping at the top and bottom of the waveform.; that will be near 12V, but not exactly. The instructions say to set the gain switch full CCW then go 5 clicks CW. Looking at the front panel photos, I read full CCW as 15dB and five clicks CW from there would be 40dB for a gain of 100. That means a 0.5Vpp waveform on the input should be 50Vpp on the output, well outside the power supply range. All this means is there is a fair amount of transformer gain on the output, maybe 10dB or more.
The output driver is a 2N3055 (that should be the one in the TO3 metal case on the heatsink) and another indicator there is some significant gain in the output tranny. The collector is the case on these, so real easy to get to). Measure the DC voltage (with no input drive) from collector of the 2N3055 (the metal case of the transistor) to ground. it should be near 12V, say +/- 2V. If significantly off, adjust the bias (R39) to get it near 12V. If you can't do that, check the values of R38, R39, and R40. On R39 check the resistance between both outer legs and from the tap to each leg. Trimpots are notorious for the wiper being open.
If everything is OK to this point, give R19 a twist with a screwdriver while applying your 0.5V sine wave. This is a gain trim. If it has no effect on the output, make sure R20 is in place and if it is, it really is 0 ohms. If everything is OK there, check R19 like you did R39 and make sure the resistance at the tap really changes when you turn the scres, then go back and double-triple check the resistors on the gain switch. Since you can back off the input signal and things look good I doubt there is a problem in the preamp (Q1, Q2, and Q3). It is either in the gain switch or the output driver. If you can rule out the driver, be very suspicious of the gain switch.
To test the gain switch, I would crank back the input to a distortion free level, then run the gain switch through each setting. You should see 5dB steps (about 1.8X) gain change with each click comparing output to input on your scope. If you don't see that, then it's either the switch or one or more of the associated resistors. I have, over the years, encountered many open resistors and mismarked resistors direct from the supplier. Now, with the uncertain origin of so many parts that sort of thing has gotten worse. I always ohm out resistors now; partly because my eyes are bad and partly because it's a lot easier to find a bad part before its installed than after. The days of factory tested resistors went away in the early 80s. Also, as a last ditch effort, ohm out the switch.
One last thing - I have looked at 7th Circle designs over the years and consider them one of the best modern designers around these days. They pay attention to things most people don't even think about, life RFI/EMI. Most of the stuff I see these days I wouldn't trust within a mile of a dimmer bank. This stuff I would take to any stage in the country. But the output tranny has DC on it. All I know about it is it's Carnhill. Trannys need to be designed for DC or they become magnetized. Ian can probably tell you more about why that's bad; all I know is it's bad. These guys are smart enough to spec the right tranny, but that is one thing I would want to be certain of before I plunk down $$$ on a preamp.

I got long winded but sometimes it takes a lot of words to explain something. My troubleshooting approach has evolved over the years and may not match yours; that's ok so long as someone figures out the problem. In this case, all prior comments point to too much gain, which suggests an issue in the gain switch since everything works with lower signal. Why not start there? Because life has taught me to check the basics: power first, always. Then, check the DC operating point. Then, if that is ok, do what your brain told you to do in the first place. YMMV.

 

[soapfoot]

Thanks for taking the time to offer your suggestions!

Just for some background, this is the seventh time I've built one of these (and this is one of three I built with this batch). I'm pretty familiar with the calibration procedure and what to expect. This is the only time I've seen this behavior.

Here are the steps that have been taken so far:

• both sides of board inspected with magnifying glass for potential shorts
• each part double-checked against the BOM and schematic to ensure all correct values are present, and all capacitors are installed in correct polarity
• Output of U1 is 23.3VDC
• Collector of Q6 shows 22.18 VDC with respect to ground
• R39 verified to trim properly; set the spec'd bias voltage starting point of 3.75V at R40 with no issue
• Tested voltages of Q1-Q6, as follows:

	C	B	E
Q1	3.993	2.488	1.877
Q2	12.77	3.993	3.342
Q3	21.53	12.78	12.11
Q4	5.28	0.922	0.321
Q5	22.09	5.27	4.67
Q6	22.09	4.65	4.04

Measurements marked in red are slightly high with respect to the specified values in the documentation, but not extremely so. But with that in mind...

• Removed Q4 and Q5; tested with transistor tester; test OK
• Removed J2, no change
• Verified signal generator output is clean
• Reflowed all joints on board again
• Double-checked each part against BOM for correct value and orientation again.
• Checked that R19 is functioning properly
• Inspected entire board with magnifying glass again for potential shorts or cold joints

Distortion is present at all gain settings, both on the stepped gain switch and all positions of R19. Stepped gain switch behaves normally at each position (in terms of adjusting level). All joints on this switch (SW1) have been reflowed twice.

I can scale the pictured waveform up or down in amplitude, but I can't make the distortion going away without reducing signal generator output level to well below 0.5Vpp. To reiterate: none of the other six units I've built exhibit this behavior under these conditions.

I've puzzled over the schematic to try locate the "loop gain setting network" as mentioned by John Roberts above, but I'm afraid I might be underqualified to determine exactly what/where that is. I did pay special attention to some areas that looked like they may relate to loop gain, but I was really only just guessing.

Out of your suggestions, the only box I haven't ticked is verifying that R20 measures zero ohms. It's definitely in place and the correct value, but I'll check that now.

 

[JohnRoberts]

Could you provide what the red marked voltages should be, either from measuring a good unit or documentation.

Nothing jumps out as obviously faulty, those are valid voltages for working transistors but perhaps not what they should be.

JR

 

[soapfoot]

Hi John, thanks.

Here's what's listed in the manufacturer's documentation:

 

[JohnRoberts]

Collector Q4 looks a little high. Does varying R39 trim affect that collector voltage?

JR

 

[soapfoot]

Collector Q4 looks a little high. Does varying R39 trim affect that collector voltage?

JR

It does indeed

 

[swpaskett]

Yeah, all very interesting. The collector of Q6 is 2 Vbe drops below the rail, telling me it is turned off, not what I would expect for Class A operation, but that is close to what the documentation says to expect. I would expect it to sit near 12V with no signal applied. So now I am baffled. Maybe JR can point us both in the right direction.
The gain setting loop is all tied up in that gain switch. There are multiple feedback loops -- some on the switch and some local -- but look closely at the SW1b and SW1c decks on the schematic; they are in the middle of it all. SW1a appears to form a voltage divider with the input bias network (R15 || R21) and does not technically set "gain" but impacts gain of the overall circuit, if I am making sense at all. (I have been out of the analog game so long I am reaching the limits of my understanding as well. JR and Abbey Road will not steer you down a blind alley like I might, though.)
One question - these voltage measurements you made - are they with no signal or with signal applied? That could possibly account for the somewhat off numbers. The documentation does not say one way or the other, but generally they would be made with no signal.
Hats off to the cats at 7th Circle, though. The kit is very well documented. You might be getting to the point at which you ohm out the board. Sometimes it's the PCB itself that is at fault.
Another idea is to ask around if someone has simulated this circuit. Then you can work at breaking the sim until you get similar behavior. OTH, 7th Circle has probably seen this before. Have you asked them?

 

[moamps]

Yeah, all very interesting. The collector of Q6 is 2 Vbe drops below the rail, telling me it is turned off, not what I would expect for Class A operation, but that is close to what the documentation says to expect. I would expect it to sit near 12V with no signal applied. So now I am baffled. Maybe JR can point us both in the right direction.

Q6 is loaded with a transformer so it is just fine.

@soapfoot
DC voltages are ok. Could you measure the AC voltage on potentiometer?
Where did you connect the measuring probe when you got the waveform from the first post?

 

[soapfoot]

@soapfoot
DC voltages are ok. Could you measure the AC voltage on potentiometer?
Where did you connect the measuring probe when you got the waveform from the first post?

Measuring probe was connected to the output XLR, CONN2, pins 2-3

All DC measurements were taken with no signal applied.

When you say to measure AC voltage on potentiometer, are you referring to R19 or R39? And do you want this measurement with signal applied or no?

Also, to answer @swpaskett question: yes, I've been in touch with Cory at SCA and he has been very helpful suggesting things to try. At this point, though, his suggestion is just to send the assembled module to him so he can have a look. It would be great to avoid that if I can, but I'm willing to do so as a last resort.

 

[JohnRoberts]

does it adjust to correct nominal voltage?

JR