Power transformer clipping?!

[jimboyogi]

Hi All,

I've just put together a SE modified 5F2-A. It's not working right. There are many questions that I will be asking, so please don't get annoyed with me

First things first though, the AC secondaries from the PT are significantly clipped on both top and bottom of the waveform when viewed on a CRO.
The PT is a new AnTek AN-IT250, with 500V CT and 2 * 6.3V secondaries. All are used in the circuit. With no valves inserted I am getting 261VAC from each of the two HT to CT, and 522VAC HT to HT. Heaters I am getting 6.2VAC on both. Slightly low but I didn't think anything of it until I put the CRO on them!

The only load operating without the valves is: 470K bleeder resistors across each of the filter caps, and 100 ohm artificial centre tap resistors from each of the output valve heater pins to it's cathode.

Does anyone know what's going on here?

Thanks for any help,

James

[MWJB]

Please specify what isn't working right?

What dc are you getting at pin 8 of the rectifier?

[Steve Conner]

First things first though, the AC secondaries from the PT are significantly clipped on both top and bottom of the waveform when viewed on a CRO.

That's no cause for concern. The mains voltage itself is often clipped straight out of the wall, because of the large amount of gear in use with switched mode power supplies. These have a huge peak current draw that squashes down the peak voltage. It's especially bad if you're in an office building with lots of computers.

And the transformer just reproduces the same waveform it gets from the mains on all its windings.

Also, the rectifier in your amp does exactly the same thing, clipping the peaks on the HT winding even more. But the effect should be minimal with the tubes pulled.

[jimboyogi]

Please specify what isn't working right?

What dc are you getting at pin 8 of the rectifier?

Hi MWJB, i'm using a full wave diode rectifier, and getting 327 VDC at the first filter cap +. When I put the CRO on this point, it shows a sawtooth AC waveform of ~ 5 V peak to peak. At the upper point of this waveform where it changes direction, there is a high frequency spike. When I insert the valves and switch off the NFB, the high frequency spike is visible on the lowerpoint of the sawtooth as well.

I was wondering why am I only getting 6.2 VAC on the heaters? I wondered if the clipping was responsible, but it looks like that is not it. As the PT has 2 * 115VAC primary windings, I have connected the 2 windings in series. But I am giving it 240 VAC, so I expected a slightly higher heater voltage than 6.3V, not lower.

A few things are not working right on the amp.

The second gain stage (1/2 12AX7 exactly as for 5F2-A) is both grid current limiting and cut-off clipping at ~ 50V p-p at it's plate, and only has a voltage gain of ~ 10. I was expecting a voltage gain of ~ 35 (no cathode bypass cap). The bias is close at 1.4V between the cathode and grid.

The output stage is with a 6CA7, and is taken directly from MerlinB's website for a SE output. The only differences are a higher HT (327V instead of 300V) and a higher OT primary impedance (I calculate 4.7K, instead of 4K). The bias is running at 20V ground to cathode.
The signal at the 6CA7 plate (pin 3) starts to clip at ~ 180V p-p, at a much lower voltage than I expected. I calculated that this is only 0.8 Watts RMS into 8 ohms. The cathode resistor is getting fairly warm as well, it is 220 ohm/5W, and should only be dissipating 20v/220 ohm = 0.09W?

When the volume controls (it has a master volume, a 500k pot in the place of the 470k 6CA7 grid resistor) are turned up, even with nothing plugged into the input, there is an effect like a phaser operating on the hum/noise, i'm guessing at less than 1 Hz. You can see this on the CRO too, it's a little glitch on the sine wave signal that slowly moves along the waveform, sometimes to the right of screen, sometimes to the left.

These are the things that I wanted to fix first!

Thanks,

James

[jimboyogi]

That's no cause for concern. The mains voltage itself is often clipped straight out of the wall, because of the large amount of gear in use with switched mode power supplies. These have a huge peak current draw that squashes down the peak voltage. It's especially bad if you're in an office building with lots of computers.

And the transformer just reproduces the same waveform it gets from the mains on all its windings.

Also, the rectifier in your amp does exactly the same thing, clipping the peaks on the HT winding even more. But the effect should be minimal with the tubes pulled.

Thanks Steve, I should have thought to check the mains voltage before getting worked up, as you predicted it is also clipped to the same degree as the PT secondary waveforms. I guess this is not the problem causing my other problems!

Thanks again Steve,

James

[Steve Conner]

What screen supply voltage, screen resistor, and output transformer are you using?

Your cathode resistor is dissipating P= V squared/R = 400/220 = 1.8 watts.

The phaser noise is probably getting picked up from something else on your bench, running at "nearly 50Hz". Say, a CRT TV or video monitor.

[MWJB]

6.2VAC on the heaters - if this doesn't rise significantly without a load on the heaters & B+, then put it down to manufacturing tolerance. I'm not familiar with the PT you describe, but I wouldn't assume that there is a problem there yet. 6.3VAC is a nominal rating, it is possible that you could see 6.1 to 6.8VAC and for anything in between to be normal.

You mention 2x115VAC primaries, do you have any other primaries to try (2x110VAC)?

327v HT is no problem, the EL34 in SE operation will handle a 100v on top of that.

The bias is not measured in volts, you should be measuring it in mA, a quick sum tells me that 20v/220ohms cathode resistor puts your plate current around 90mA (as you have already deduced). 90mA*307v = 27.8W plate disipation...this is a lot. I would go up on the cathode resistor value to 330ohms at least, at a 10W rating, B+ voltage will rise but not to dangerous levels. I'd aim for more like 22-23W plate dissipation max. What dc current is your PT rated for, what 6.3VAC caurrent is your PT rated for?

In short you want more volts & less curent at the 6CA7.

6CA& cathode resistor will get warm, keep the bypass cap physically away from it, cap should be rated at 50v min, preferably 100v.

I think that your voltage gain expectation at V1b is optimistic.

Please post schem & pics.

[Wilder Amplification]

In regards to your expected gain factor...if it were a normal cathode bypassed preamp stage you probably would have a bypassed gain of about 35dB. However that stage is getting negative feedback from the amp's output. So on your expected gain I side with MWJB on the fact that expecting a 35dB gain factor on that stage is a bit optimistic.

[ThermionicScott]

Without the cathode bypass cap, the gain is cut roughly in half, isn't it? A single-ended EL34 ought not to need much drive anyway. I use a 12AY7 in my 5F2A to tame it a bit.

[jimboyogi]

Thanks all for your comments and questions, I'll try to answer each of them.

Firstly Steve, screen dropper is 470 ohm, dropping 8V to 319V, followed by a 470 ohm screen stopper dropping 5V to 314V at the screen.
The OT is a 100V line matching transformer, rated at 40W, with multiple taps for primary and secondary. I am running it on the 5W primary tap, and putting an 8 ohm speaker/dummy load on the 4 ohm sec. tap. This was supposed to give me a reflected load of 4K primary. When I measured the voltage ratio across the OT I calculated that I was getting 4.7K primary load.

MWJB, there are no more primary taps on the PT.
With the bias, as you say I am getting 90mA cathode current, 10.6mA screen current (5V/470ohm), leaving around 80mA plate current. 80mA*307V = 24.6W, which is still very close to the 25W max. plate dissipation. I will try a larger cathode resistor, and the bypass cap is rated for 100V.
The PT is a biggie for the job, 200mA HT and 3A*2 for the heaters.
I'm working on putting up a schematic and pics.

Jon, I was expecting a voltage gain of ~35, not sure what that is in dB. I actually did these measurements with the NFB off to help simplify how it was behaving. The figure of voltage gain of ~35 is taken from MerlinB's book, first chapter, where he shows the exact same circuit (100Kohm plate load, 1.5Kohm cathode resistor, 280V HT, 1 Mohm grid resistor) but not accounting for the output impedance loading. In my circuit this is a 500K pot (master volume), which at 5* the plate load should not load the circuit too much?

ThermionicScott, this build has a master volume, and the guy I'm building it for wanted a fair bit of pre-amp drive on tap.

V1b is biased at 1.4V, and I'm getting 4.8V out of V1a from a 0.1V signal input, so I should be able to get a fair bit of overdrive here. The problem is that V1b doesn't seem to be behaving as I expected. From a HT of 290V shouldn't I be getting more than 50V p-p voltage swing before clipping/grid limiting?

Thanks again for all of your help people,

James

[Wilder Amplification]

Jon, I was expecting a voltage gain of ~35, not sure what that is in dB. I actually did these measurements with the NFB off to help simplify how it was behaving. The figure of voltage gain of ~35 is taken from MerlinB's book, first chapter, where he shows the exact same circuit (100Kohm plate load, 1.5Kohm cathode resistor, 280V HT, 1 Mohm grid resistor) but not accounting for the output impedance loading. In my circuit this is a 500K pot (master volume), which at 5* the plate load should not load the circuit too much?


James

Voltage gain of 35 translates to 30dB (log 35 x 20 = 30dB)

However, according to my math you should be getting a voltage gain of about 27 with that 500K pot as your load. Is this about what you're getting?

[Steve Conner]

The OT is a 100V line matching transformer

Oh... Those don't have a gapped core, so they're not too good for single-ended use. That might explain the low power output.

[MWJB]

"The figure of voltage gain of ~35 is taken from MerlinB's book, first chapter, where he shows the exact same circuit (100Kohm plate load, 1.5Kohm cathode resistor, 280V HT, 1 Mohm grid resistor) but not accounting for the output impedance loading."

With those values and 280v feeding the 100K plate resistor, I'd expect plate voltage to be around 190v. Which in turn suggests 0.9mA current draw, as backed up by your 1.4v cathode voltage. All seems to check out fine...vol & tone control are fully clockwise when AC voltages are taken?

With a low voltage SE amp, irrespective of whether you have a 6V6 or 6CA7, 6L6 or whatever, you're probably going to find that a degree of clipping sets in earlier than you are expecting...I would measure actual output at normal playing settings & work back from there. If the amp makes 6 or 7W, I wouldn't expect it to be clean W.

If you are only making 0.9W at the speaker/dummy load, irrespective of how clean the signal, then you have an issue.

[ThermionicScott]

Soundwise, I don't know how this would play out, but it is possible to extract more gain from your preamp stages. If you set up the second gain stage and NFB more along the lines of a BF Champ (with NFB applied to the bottom of the bypassed cathode resistor), that'll help. For the first stage, you can try a larger plate resistor (like 220K or even 470K), but you'll then need to tweak the cathode resistor upwards a little.

- Scott

[jimboyogi]

Thanks for your continued help

Jon, your calculation is very good! I re-measured the gain on V1b for a clean sine wave and got a voltage gain of 28.2 with no NFB.

Steve, please inform me about the problems with non-air gapped core OT's for SE use. I've read that due to the constant DC current through them the core can saturate, rolling off the low frequency response. I've also read that you can help to improve their low freq. response by using one with bigger iron core, that is why I used a 40W rated OT in a 10W amp. Only cost $4 more than a 20W one If they really won't work then I'll have to replace it, but they are so cheap I thought it was worth a try.

MJWB, all of your numbers match what I measured, but when I put the CRO on the sine wave signal I saw grid current limiting way before cut-off clipping with this bias, it seemed nowhere near centre-biased. I was really dis-liking the amps inability to give a decent volume clean sound.
I ran a load line for V1b with a colder bias of ~ 1.7V, and then subbed a 2.2K in place of the 1.5k cathode resistor. Now on the CRO grid and cut-off are almost completely symmetrical, and the amp now has much more pre-amp headroom, and to me a sweeter clean sound. The pre-amp overdrive is now a bit grittier, but that's OK!
I haven't re-measured the voltage at the speaker output yet. Can you confirm what the formula to convert speaker p-p voltage to Watts RMS is?
I've got: p-p V/2.8 = RMS V, P = RMS V^2/R of speaker/dummy load.

ThermionicScott, thanks for these tips. I may yet use them.


Another question re. hum reduction. When I had another look at the 12AX7 data sheet, I saw that a note that the triode on pins 6,7,8 gives the lowest hum. I have mine wired with the input into triode pins 1,2,3. Is there actually a discernible difference to the hum level, and should i re-wire mine?

Thanks again,

James