5F6-A ugly distortion - Direct coupling question

[jmaf]

Hi Steve,

I guess, yes to some degree it might be, but the OP seems to be describing something that isn't typical. For some reason his preamp voltages appear to be ~35v down, this might be just enough to underbias the preamp tubes & when hit with a large signal they suffer. It's not unusual to use lower mu tubes in V1 with humbuckers, but the amp should certainly be clean with a 12AU7 in V1, if not with a 12AY7 in V1.

It now emerges that this is a 1 channel amp, if the OP has wired the "spare" triode in series, then a much larger cathode resistor (10K unbypassed) is not uncommon for the 2nd triode (1st triode needs to be corrected to 1.5K). If he hasn't wired in the spare triode at all, then voltage at V1 should be even higher. If V1 triodes have 820ohm at each triode, rather than shared, then that could explain the low voltage and the ugly distortion.

The Selmer Corvette is a very low voltage amp, with no attenuation between the 2 halves of the 12AX7/ECC83...it probably needs those 5K to get any clean at all (Kalamazoo Model 2 also used 4.7K cathode resistors, but had the more typical vol/tone placement between the 12AX7 triodes). The 5F6A with a 12AY7 in v1 should be on par with a Super Reverb clean-wise.

Plus the biggest factor in this, is that the amp in question is not a proven, road tested amp. The schematic might be, there are tens of thousands of 5F6A that are, but this is one amp that we have never seen, built to a schematic (unseen) which is just a bit of paper that does not supply enough information to do the job. There's a lot of scope to screw up layout...Marshall did it with the 1962 RI, as you know from the Rattling PRRI thread, Fender seem to have done it with that amp too...it's too easy for a builder unfamiliar with the foibles of that particular amp to make an error. I know how easy it is, because I've done it too in the past (all values & voltages check out...the amp just never works properly, sub all the parts you like, if you just put them back in the wrong place, then it never will)...as have many of the people that make suggestions as to how to cure such problems.

I'm not precious about changing values in any amp, to suit the player, it accounts for most of the work I do...I just don't do it to mask a malfunction, make sure that everything works and you are at square one, before modding.

If changing that resistor was the right thing to do, I'd be happy. This one did not sound like overdrive to me... I've taken note of your suggestions, and I'm heading back to shop. I don't have a PC there, so I'll test and see. I will take better measurements to provide, I realise I didn't give much for you to work on.

Spare triode is planned for a reverb return, not in the circuit yet. I've tested it, it's wired, it works ok, but it's not in this circuit we're discussing yet. So for all practical purposes, it's a loose triode.

[Steve Conner]

The Selmer Corvette is a very low voltage amp, with no attenuation between the 2 halves of the 12AX7/ECC83...it probably needs those 5K to get any clean at all (Kalamazoo Model 2 also used 4.7K cathode resistors, but had the more typical vol/tone placement between the 12AX7 triodes).

I'm always tempted to try shifting the volume and tone controls to between the two triodes. Mine is running considerably higher voltages than stock now, anyway.

[txstrat]

If V1 triodes have 820ohm at each triode, rather than shared, then that could explain the low voltage and the ugly distortion.

That's what I was referring to in my first post. I think I somehow KNEW it was a one channel amp. (kidding)

[jmaf]

I finally took note of all voltages - thanks for your patience here in advance.

Here I go:

Power Supply
My power supply is +B -> 1K resistor (+B1) -> 4k7 resistor (+B2) -> 2k2 resistor (+B3) -> 2k2 resistor (+B4)
Caps for each pi filter are 220uF totem pole(total 110uF), 40 uF, 20 uF, 20 uF, 20 uF

+B = 465 VDC
+B1 = 444 VDC
+B2 = 397 VDC
+B3 = 381
+B4 = 370

BIAS 6L6 1 = -45
BIAS 6L6 2 = -45

G2 6L6 1 = 453
G2 6L6 2 = 453

Plate 6L6 1 = 455
Plate 6L6 2 = 455

Plate V1 (12AX7) = 326 VDC
V1 Cathode = 3,69 VDC
(Remember this one has a 10K cathode resistor, it's cold as can be.)

V2A Plate = 191,03 VDC
V2A Cathode = 1,32 VDC

V2A Plate = 367,65 VDC
V2A Cathode = 193,30 VDC

Inverter
V3A Plate = 251
V3A Grid = 29,64
V3B Plate = 248
V3A Grid = 31,90
V3A/B Common Cathode = 50

(Here a question arises: from ground, there should be a 20 V difference between cathodes and grids here, but when I measure straight at the pins, I get -1,15 volts precisely....how can this be? I thought I had finally found the problem...not so.)

Resistance from phase inverter grids to ground: 1M
Resistance from PI cathode to ground: 18K

AC Measurements]
Using peak VAC from humbucker output: 300mV sin wave

All Voltages VAC PEAK on oscilloscope.

V1 Grid: 300 mV
V1 Plate: 1,62

(Volume set on 2, where strange distortion arises w/ guitar ear test.)

V2A Grid: 60mV
V2A Plate: 2,32

V2B Grid: 2,32 (direct coupled)
V2B Cathode: 2,36 (some gain???)

PI Grid 1 = 1,20 VAC
PI Grid 2 = 1,18 VAC

6L6 Grids: Both at 10,8 VAC
6L6 Plates: Both at 64,00 VAC

Do these seem ok to you? I've banged my head with this amp, the distortion is horrible, fuzzy, buzzy, etc. It doesn't seem like overdrive. Any ideas extremely appreciated. Thanks again for your help.

[pdf64]

Have you got access to an oscilloscope?

'Here a question arises: from ground, there should be a 20 V difference between cathodes and grids here, but when I measure straight at the pins, I get -1,15 volts precisely....how can this be?'

Your meter will have an input resistance of about 10M, so when measuring from ground, it's pulling the grid down. When measuring from the cathode, it's pulling the grid (very slightly) up.

[MWJB]

Why the 1K resistor between the OT CT & screen supply (B+1)? If you haven't used a choke check out the DCR of a typical choke, more likely to be in the region of <250 or 330ohms.

-45vdc at grids? Your negative voltage should be 11% of the power tube plate voltage as a starting point (-50v) at the junction of the 220K bias feed resistors. Fine tune plate current from there, 35mA per tube will be fine.

B+3/B+4 at either 381 or 370vdc your plate voltage on V1 should be in the region of 250vdc with 1640ohms per triode, I had assumed wrongly that you had the 10K at another stage, you must have had a wrong value somewhere to have been getting 190vdc on V1 plates as before (unless you had a single 820ohms on V1 used triode, looks like you might have as plate voltage is half B+? Should be nearer 2/3 of B+ supply).

The AC voltages don't mean a geat deal because you haven't set the amp like the 59RI schem, tubes amplify by different ratios at different B+ and depending on how hard they are pushed. V2 at x40 from grid A to B cathode looks OK. set the amp as per the RI schem. Your PI looks to be slightly down but that could be down to how you have things set, unless you have the 10K tail resistor instead of the '59 6.8K tail?

Any pics of the circuit board, pot & tube socket wiring?

[jmaf]

Have you got access to an oscilloscope?

'Here a question arises: from ground, there should be a 20 V difference between cathodes and grids here, but when I measure straight at the pins, I get -1,15 volts precisely....how can this be?'

Your meter will have an input resistance of about 10M, so when measuring from ground, it's pulling the grid down. When measuring from the cathode, it's pulling the grid (very slightly) up.

My DVM would shunt a 1M resistance by 10%(10M), yet cause such a big discrepancy in the reading? I can read all other grids on the circuit, and other high impedance circuits with no such problem. Why does it distort the reading so much on the PI?

Why the 1K resistor between the OT CT & screen supply (B+1)? If you haven't used a choke check out the DCR of a typical choke, more likely to be in the region of <250 or 330ohms.

-45vdc at grids? Your negative voltage should be 11% of the power tube plate voltage as a starting point (-50v) at the junction of the 220K bias feed resistors. Fine tune plate current from there, 35mA per tube will be fine.

B+3/B+4 at either 381 or 370vdc your plate voltage on V1 should be in the region of 250vdc with 1640ohms per triode, I had assumed wrongly that you had the 10K at another stage, you must have had a wrong value somewhere to have been getting 190vdc on V1 plates as before (unless you had a single 820ohms on V1 used triode, looks like you might have as plate voltage is half B+? Should be nearer 2/3 of B+ supply).

The AC voltages don't mean a geat deal because you haven't set the amp like the 59RI schem, tubes amplify by different ratios at different B+ and depending on how hard they are pushed. V2 at x40 from grid A to B cathode looks OK. set the amp as per the RI schem. Your PI looks to be slightly down but that could be down to how you have things set, unless you have the 10K tail resistor instead of the '59 6.8K tail?

Any pics of the circuit board, pot & tube socket wiring?

The 6L6's are at 30mA per tube right now. I checked with an oscilloscope, I don't see any crossover distortion at all.

The 1K resistor is there because the multisection can cap is rated 450VDC and I wanted to be safe with the peak voltage spike during turn-on. It's a 5 Watt wirewound.

Three days into this novel already! Any other ideas?

[MWJB]

You should have 50v headroom over the cap voltage rating, ideally. So is there a choke or not? If there is, where is it's filter cap for the screen supply?

"Three days into this novel already! Any other ideas? " Lets just work through what we have for now...put in the correct value cathode resistor at V1 and try and ascertain why the voltage at the plate does not meet expectation. We still need pictures please.

[jmaf]

You should have 50v headroom over the cap voltage rating, ideally. So is there a choke or not? If there is, where is it's filter cap for the screen supply?

"Three days into this novel already! Any other ideas? " Lets just work through what we have for now...put in the correct value cathode resistor at V1 and try and ascertain why the voltage at the plate does not meet expectation. We still need pictures please.

OK, thanks. There is no choke, the power supply chain described is it all.

I won't be at the shop today, monday I'll get pictures taken and return the V1 to original value.

[MWJB]

"OK, thanks. There is no choke, the power supply chain described is it all." Then how come your screen grid voltage is 9v higher than the B+ supply at the screen node (B1)? Have you installed discrete screen grid resistors (470ohm-10000ohms) feeding pin 4?

[Steve Conner]

My DVM would shunt a 1M resistance by 10%(10M), yet cause such a big discrepancy in the reading? I can read all other grids on the circuit, and other high impedance circuits with no such problem. Why does it distort the reading so much on the PI?

The input impedance of a LTPI isn't 1M. The 1M grid resistor is bootstrapped to the cathode, and that increases the effective "resistance" between that point and ground. I think it increases it by a factor of gm times the tail resistor, or something, which is about 20 to 50.

For the same reason, I guess the LF rolloff due to the PI input coupling capacitor is much lower than you'd think.

[jmaf]

"OK, thanks. There is no choke, the power supply chain described is it all." Then how come your screen grid voltage is 9v higher than the B+ supply at the screen node (B1)? Have you installed discrete screen grid resistors (470ohm-10000ohms) feeding pin 4?

G2 6L6 1 = 453
G2 6L6 2 = 453

Plate 6L6 1 = 455
Plate 6L6 2 = 455

Plates are 2 VDC higher than G2's. +B is +10 Volts with relation to plates, due to quiescent current through the winding resistance.

Yes, 470 ohm resistors in place.

Edit: I see what you mean about +9 Volts. From +B I take a tap, decouple it via a 20uF totem pole and it goes into screens (G2) via 470R resistors.

[jmaf]

The input impedance of a LTPI isn't 1M. The 1M grid resistor is bootstrapped to the cathode, and that increases the effective "resistance" between that point and ground. I think it increases it by a factor of gm times the tail resistor, or something, which is about 20 to 50.

For the same reason, I guess the LF rolloff due to the PI input coupling capacitor is much lower than you'd think.

Thanks for the insight. In your opinion should I lower the cap value to increase the rolloff frequency and see if the ugly distortion goes away?

[pdf64]

Best to get the amp working correctly before modding.
Can you get the ugly distortion with a sine wave input?
If so you should be able to see where it's being introduced by checking with your scope. Take a photo and post it if possible.

[Scott_F]

Grid stoppers. 1K5 on each pin 5 power tube for what it's worth.