Mojotone Custom 100 (Hiwatt 103 copy) Build Question

[ThreeChordWonder]

So I've built this Mojotone Hiwatt 103 copy, and after a few bumps in the road, I've pretty much got it working.

Except...

When I strum lightly, no problem. When I strum a little harder the sound breaks up and I get this horrible arc-ing sound.

I think I've tracked down the source, if not the cause, of the problem.

I think the inverter tube V4 plate voltages are out of balance, one too high, the other too low.

The plate voltages on the inverter tube don't match the circuit diagram.

V4 plates are fed off a 457 volt feed from the power rail, just "downstream" of the 1K 2W resistor. That voltage is measuring 464 volts, which is close enough.

The cathode voltage is pretty much spot on at 87.5 volts vs 85.3 shown on the wiring diagram.

I'm attributing these slight overages to out mains being a bit higher than 110 volts this morning.

The V4A plate is fed through an 82K resistor. The plate voltage should be 307 volts. I'm seeing 460 volts, just a 4 volt drop, which to me suggests the plate current is too low.*

The V4B plate is fed through a 91K resistor. The plate voltage should be 21 volts. I'm seeing 172 volts, which to me suggests the plate current is too high.*

* Ohm's Law V=I.R also means delta-V, or the voltage drop across a resstor R also equals I.R so if there isn't enough current, I, flowing, there won't be enough delta-V either.

Questions are;

1. Are my voltages correct (unlikely) or is the circuit diagram right?
2. What could be causing this? It's pretty simple part of the circuit, after all.
3. Should I try swapping the resistors over?
4. Any other suggestions welcomed.

[martin manning]

V4A is hardly passing any current at all since its plate is nearly equal to the supply. Bad tube or bad socket maybe. I don’t know why the cathode voltage is so close with only one triode drawing current though. The plate voltages on the schematic look wrong, they should be closer to each other than the values shown, maybe within ten volts of each other.

[ThreeChordWonder]

I tried a different ECC81, made no difference. New socket might be worth a try.
Thanks.

[martin manning]

Could also be a wiring error or wrong component value driving the grid voltage of that first triode low and putting it into cut off.

[ThreeChordWonder]

I've been over the wiring a dozen times and I checked each component with a multimeter before it was soldered in. Then I checked all the joints for continuity.

Also, there's only two resistors between the HT off the rectifier diodes and the 82k / 91k split, and only one 47 nF between each pair of output tubes and their individual 22k grid resistors. All of those were checked too.

Finally, all the voltages upstream of the 82k /91k split are spot on.

The only thing I haven't checked is the socket. I guess a bad pin receptacle there could be the culprit. No juice flowing through the V4A side enabling too much juice to flow through the V4B side, maybe?

[martin manning]

The second triode is drawing a lot of current, and that is making the cathode voltage look ok. The plates should both be around 320V.

[ThreeChordWonder]

The second triode is drawing a lot of current, and that is making the cathode voltage look ok. The plates should both be around 320V.

Thanks.

It's what I figured.

Once I've got them in the same ballpark, I guess I can play with the 82k / 91k pairing to get them closer.

I've ordered some new sockets, $10 on Amazon, and we'll see what a new socket does, or does not, do.

[martin manning]

Once I've got them in the same ballpark, I guess I can play with the 82k / 91k pairing to get them closer.

Those resistors differ in value to balance the outputs.

[ThreeChordWonder]

Yes.

Get them closer to the target voltages, not necessarily equal.

[ThreeChordWonder]

A few photos.

Bear in mind I'm not a pro with a workshop, and that this is my first build, so be kind

I fitted a 5k pot so I could adjust the power tube bias without swapping out resistors. I also fitted 1 ohm resistors as the ground wires for the power tubes, so I can measure plate currents by measuring the voltage drops across those resistors. I also fitted pin 5 probe points so I can measure bias voltages more safely.

[cdemike]

Yes.

Get them closer to the target voltages, not necessarily equal.

Pretty build! I like the bias pot bracket.

I don't think they'll necessarily be equal unless the plate resistors are equal, and that's actually best -- as Martin said, the reason for selecting mismatched resistor values in a long tail phase inverter is to balance output, not necessarily plate voltage. This simulator shows the impact of using different values in that position (https://www.ampbooks.com/mobile/amplifi ... iled-pair/), and Uncle Doug has a video I liked explaining why the output is unbalanced (https://www.youtube.com/watch?v=U6By31V9fDo). So while the voltages won't be equal, each triode will be delivering comparable signal amplitude when mismatched due to the asymmetry of the the signal path in a LTPI between each bank of output tubes. Consider this: each side gets B+ voltage from the same node and presumably the triodes are fairly similar to each other in any given 12AT7, so if the design required equal plate voltage on each triode, Dave Reeves, Leo Fender, Ken Bran, and other notable amp designers who used LTPI's would have used equal plate resistors.

[ThreeChordWonder]

I always thought balanced output was the important thing, not necessarily input - you want your "push" and "pull" phases to be more or less equal on the output side.

Once I get the voltages about right I can experiment a little, say upping the 91k to 100k or dropping to to 92k, and / or putting a 5k in series, whilst the 82k could go up to a 91k or down to a 75k, or something in between with an incremental resistor in series.

What I've seen from the build thus far, and current problem aside, is that the Mojotone build pretty much delivers the right voltages, however. I'm using 1% metal oxide resistors generally, so my build tolerances are quite high.

I have an oscilliscope arriving in the next day or two which will help.

[ThreeChordWonder]

I like the bias pot bracket

Sheet (or should that be sh#t) metalwork at its finest, right?

[cdemike]

Sorry, let me clarify. You're correct that balanced output from the PI is the design criterion. I meant if the plate resistors' values were identical, the output from the PI would be uneven as it feeds into each bank. So it sounds like we're on the same page about that. The oscilliscope will be helpful -- you'll find that if you have the plate resisitors on the PI spec'd identically (like on an AC30, for example) you'll get uneven outputs from the PI feeding each bank of power tubes. That's because there's asymmetry in the signal path from the master volume recovery stage and each output section side. All things (including plate voltage) being equal, plate resistance tends to be proportional to the amount of gain you can expect from it; higher plate resistance, higher expect gain. Obviously that gets more complicated since higher plate resistance also means less plate voltage given identical B+ supplies, but dual triodes usually don't draw a ton of current, so sparing extreme value changes you'll usually see somewhat similar voltages with incremental tweaks in plate resistor values. In other words, the effect resistance has on gain generally outweighs the opposing impact resistance has on plate voltage and thus gain.

I should have mentioned this earlier, but have you tried wiggling V4 around in the socket to test for loose connections in the sockets? I had a bad socket in a previous build that I was able to make work by gently bending the lead from one of my tubes to make better contact. You can also do continuity tests without the tube in to see if there's a broken lead inside the socket or something similar. Forgive me if you've already done this, but have you tried the chockstick test to double check the solder joints on V4? If you have a spare 12A_7 lying around, you may also want to try putting that into V4 to see if you potentially have a dud 12AT7.

[ThreeChordWonder]

I haven't tried "prodding" the receiving end of the socket, and I'm worried that if I do I'll end up making the sockets bigger, causing more problems. The sockets themselves, with the cover bases I've fitted, don't allow much "wiggle", unfortunately, and in any case this old engineer doesn't like stuff you have to wiggle or clout like an old TV to make things work right.

My solder joints were checked with a multimeter. Just to be sure, I made sure I put the probe on the socket contact, not the wire or the solder joint.

I've already tried both a balanced ECC81, and just for luck a spare ECC83. They made no difference.