I've got an audio signal generator so I will follow that advice.
Someone did mention fitting 47pF across the feedback resistor, I've got them if necessary.
I contacted the seller of the boards and they tell me that they have sold many of these boards and their customers are all very happy.
Someone did mention fitting 47pF across the feedback resistor, I've got them if necessary.
I contacted the seller of the boards and they tell me that they have sold many of these boards and their customers are all very happy.
They're nice-looking boards and probably work fine with a little tweaking. You may need more than 47pF, more like 100 or more, but you can experiment to see. I still think C3 could be increased to 200pF to give the Hammonds a fighting chance by shelving the HF response a bit lower. And 22K for the feedback resistor isn't actually giving you much feedback, about half what a typical Williamson would want, which is 20dB. 10K to 12K would be more appropriate for a 6.6K output transformer. But you would want to take it slow and keep an eye on the scope to make sure it doesn't oscillate as you increase the feedback.
To determine the feedback level, disconnect the feedback loop from the cathode of the 12AU7 and ground it. Feed a 1K sine wave into the amp until you get 5 volts output. The reconnect the feedback loop, use the same amount of input and adjust the feedback resistor until you get .5 volts output. Less than that is too much feedback, more is not enough. Hope this helps.
To determine the feedback level, disconnect the feedback loop from the cathode of the 12AU7 and ground it. Feed a 1K sine wave into the amp until you get 5 volts output. The reconnect the feedback loop, use the same amount of input and adjust the feedback resistor until you get .5 volts output. Less than that is too much feedback, more is not enough. Hope this helps.
The schematic loosely refers to a 5.5K OPT, mine are 6K.
The Hammonds are rated 30Hz - 30KHz at 40W.
The Hammonds are rated 30Hz - 30KHz at 40W.
6k OPT is a little higher than the design value, but that just trades off power for improved open loop performance (linearity and damping). The downside is that the self resonant frequency of the transformer tends to come down. It might be above 30k, but exactly where no one knows unless you measure it.
My back of the envelope calculations put the open loop gain at about 120 (15 x 16 x 0.5). That’s not crazy high. With the feedback values in the schematic, loop gain is -2.5. That’s less than 10dB of feedback, so the OPT’s self resonance isn’t likely to get in your way. Maybe a cheap E-bay POS might give you fits, but Hammonds are better than that. You can buy better still, but I’m not sure you want to spend that kind of money.
Get the open loop gain up around 1000, and then you’ve got to monkey with it and you’ll see what all the hubbub is about. High mu triodes, or even pentode input stages can get it up there. Not 12AU7’s. Especially with pentode connected outputs, that give you 3-4X as much gain. That’s not a newbie project unless you’re following an exact set of plans, including the construction details.
My back of the envelope calculations put the open loop gain at about 120 (15 x 16 x 0.5). That’s not crazy high. With the feedback values in the schematic, loop gain is -2.5. That’s less than 10dB of feedback, so the OPT’s self resonance isn’t likely to get in your way. Maybe a cheap E-bay POS might give you fits, but Hammonds are better than that. You can buy better still, but I’m not sure you want to spend that kind of money.
Get the open loop gain up around 1000, and then you’ve got to monkey with it and you’ll see what all the hubbub is about. High mu triodes, or even pentode input stages can get it up there. Not 12AU7’s. Especially with pentode connected outputs, that give you 3-4X as much gain. That’s not a newbie project unless you’re following an exact set of plans, including the construction details.
wg_ski, I think you have an inaccurate estimate of output stage/OPT gain of 0.5 - I'd suggest it's circa 10x more - making open loop gain of circa 40dB.
Even for 5.5K that's a very high value feedback resistor. Yours are 6.6K. The typical Williamson UL feedback resistor would be around 10K. You are running triode so it would be even lower, maybe 7.5K or so.
Without the proper amount of feedback, the control of the amp will be poor and your gain will be very high. But try it as is and let us know how it works! It can't hurt to have *less* feedback. 🙂
18 volts peak out at 22 watts if you take feedback off the 8 ohm tap. Square root of two more off the 16, but you can’t access 8 and 16 simultaneously on the less expensive version of the Hammonds. Takes 34 to drive g1 in triode. That’s a gain of about 1/2. Can’t get more than mu out of a triode stage in the front end. I usually get between 14 and 18 out of an SN7, depending on what I’m doing with it.
I'm not sure why you think there isn't enough open-loop gain. Except for the fixed bias, it's a bog-standard Williamson circuit designed for 20dB feedback.
wg_ski, sorry I misinterpreted your post. Yes, the open-loop gain to speaker is circa 90-120 (ie. about 40dB). With 20dB GNFB, the original Williamson sensitivity was about 1.4-1.6V for 11.3V on an 8.5 ohm speaker for 15W.
If the 12AU7 valves are altered then all the HF roll-offs typically change for the worse.
I haven't tried to interpret the information notes in chinese to see what they say about feedback level or input sensitivity or output power rating.
If the 12AU7 valves are altered then all the HF roll-offs typically change for the worse.
I haven't tried to interpret the information notes in chinese to see what they say about feedback level or input sensitivity or output power rating.
Last edited:
Well, I've got it up and running with no flames this time.
Bias is now limited to between -33 and -50V.
Bias is adjusted to a stable 40mA.
Now, I've got no input pot on mine as the pre-amp does the attenuation.
The input grid is shunted to ground via R1 (100K).
With no input leads connected there is a fair amount of mains hum which disappears if I short the inputs to ground.
There is no oscillation and it seems stable.
Bias is now limited to between -33 and -50V.
Bias is adjusted to a stable 40mA.
Now, I've got no input pot on mine as the pre-amp does the attenuation.
The input grid is shunted to ground via R1 (100K).
With no input leads connected there is a fair amount of mains hum which disappears if I short the inputs to ground.
There is no oscillation and it seems stable.
I removed the biasing components from the PCB and have mounted them on the chassis.
R26 has been replaced with 6K8. The pots replaced with 10K and an additional 10K resistor to 0V.
This gives a safe bias range of -33V to -50V.
R26 has been replaced with 6K8. The pots replaced with 10K and an additional 10K resistor to 0V.
This gives a safe bias range of -33V to -50V.
Mains Earth is only connected to the chassis and not to the PCB or any part of the circuit.
The input leads are shielded with the screen connected at the PCB end. The input sockets are insulated as the frame is wooden.
The only 0V connections are the 0V returns from the OPTs, these are 2.5mm and are connected to the central earth point.
The input leads are shielded with the screen connected at the PCB end. The input sockets are insulated as the frame is wooden.
The only 0V connections are the 0V returns from the OPTs, these are 2.5mm and are connected to the central earth point.
I would put 10R and 100nF between mains earth and the board earth or something similar. May help.
My first impression after only a few minutes of use is that the bass is much more solid than my Aleph J.
The 10R will stop the board developing a high floating ground voltage wrt mains earth which may give you pre-amp a bit of a shock.
I'll add the ground lift tomorrow, the finished project is incredibly heavy. It's got 6 x transformers in it.