Tube Amps / Music Electronics
For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

Search for:  Mode:  

Open Source Power Tweeking

3/18/2004 12:12 PM
Steve Conner
Open Source Power Tweeking
Carl Gigun asked if I could post this into a new thread. OK here goes.  
"I've got some MTP2N60"
2 amp 600 volt? Even better, the IRF820 is only rated at 500 volts, which is a bit too close to B+ for comfort.  
"I don't understand how the circuit works from looking at it though, could you explain a little more? "
OK... I've reposted the schematic below...  
Q1, Q3, R1, R2, D2, D3, and D4 are an adjustable current source. It passes zero current when the pot R2 is set to zero, and 1 mA when R2 is set to full. This current is sunk into the collector of Q3, and the same current is sourced out the collector of Q1. (It's a funky double-ended current source :))  
The current from Q3's collector goes through M2, which is just a "cascode" to isolate Q3 from the high voltage of the B+ rail. D1 supplies the bias voltage that M2's gate needs to make it work. (Of course, I could have made Q3 a 600 volt transistor, and left M2 out. But it's hard to get hold of high gain small signal bipolar transistors with a 600 volt rating)  
Still following that 0 to 1mA adjustable current, it then goes through R4 to the B+ rail. When the current is set to 0, the voltage drop across R4 will be 0. (and the junction of R4 and M2 drain will sit at B+) When it's set to 1mA, the volt drop will be 400V. (and the junction of R4 and M2 will sit at 400V less than B+) So now we have our adjustable voltage. M1 is a source follower that buffers this voltage to give the required current to drive the amp's B+ rail.  
Meanwhile, the current that came out of the other end of the current source is doing the same thing for the power tubes' bias voltage. A cascode isn't needed here because the bias voltage is low. And I used a bipolar (Q2) as the output buffer, because it only has a 0.7 volt drop. A MOSFET would have a bigger drop, that would be less stable with temperature.  
And now, a word from our sponsors:

3/19/2004 7:10 PM
Carl Gigun

Thanks for explaining it Steve, I see how it works now. It's a pretty cool idea, a current source working against a pair of resistors to set the reference voltages for the buffers. The bias trimmer is interesting, it changes the rate the bias scales at? If that's right, you could adjust it so a 50% reduction in B+ voltage meant only a 40% change in bias voltage etc? I imagine that would come in handy for fine tuning the supply. And there's nothing like this in TUT2 you'll be happy to hear ;)  
The MTP2N60 does have 600v and 2a max ratings, but only 50w dissipation. I could cacscode two of them I guess. I'd like to be able to scale the power all the way down in a 2 tube amp.  
-Carl Gigun
3/19/2004 8:35 PM

There are things similar to this in SSH, Carl.  
Steve: The circuit needs two extra components or you will really hate dialling that pot. First, you need a cap across the power control or the DC noise will sound like an earth-quake everytime you move it. That's in TUT2. Second, you need a safety resistor across the bias trimmer. This will keep the bias voltage from dissappearing or defaulting to a low value if the pot opens. That is in TUT.  
Actually, there's a third point: the mosfet chosen won't be up to the task unless you cool it with liquid nitrogen. Better to use the larger format IRFP-series. They also have half the thermal resistance than a TO-220, just from the case design. You also get much higher power ratings and should err on the high side. IR's mosfets - and most from other manufacturers - are optimised for switching, NOT linear applications. To this end, die sizes are quite small and become a hurdle to heat management. This is covered in TUT5.  
Just as a note: Although this method hasn't been patented for use as a power control, it has been used in tracking regulator applications, which of course is what this is. It was also used as an audio power amp designs back in the 1980s.  
Carl, there should always be a way to adjust the basic balance of B+ and B-. Then, the bias pots for each tube work properly and the amp scales well.  
TUT5 will be out in a couple of weeks, as it goes to print the end of next week. It has a complete step-by-step amp construction that incorporates Power Scaling and Sag control. This represents one way to do these things, as there are lots more. The dual-pot approach was mentioned in TUT2, and described in detail in SSH as it is the subject of a patent.  
Anyone who wants to use Power Scaling, in fact, can. They just have to contact me and provide proper acknowledgement to the source of the information. Powr Scaling and Power Scale are trademarks so their use must be approved by myself. Taking the honourable route also gets you tech support.  
Have fun  
Kevin O'Connor
3/19/2004 9:27 PM

"TUT5 will be out in a couple of weeks, as it goes to print the end of next week. It has a complete step-by-step amp construction that incorporates Power Scaling and Sag control."
Fixed bias implementation?
3/20/2004 12:14 AM
Carl Gigun

There are things similar to this in SSH, Carl.

Just my luck, that's the only one I don't have! :)  
Thanks for the tips Kevin, and congrats on finishing TUT5.  
3/20/2004 12:43 AM
"First, you need a cap across the power control or the DC noise will sound like an earth-quake everytime you move it. That's in TUT2."
I'm not arguing the rumble point, however, that's not quite fair, Kevin. Figures 2-42 aned 2-43 are pulling off of the AC to create a DC reference voltage. I don't see "a cap across" Steve's R2 being the same. I'm not sure what you mean, ftm.
3/20/2004 2:28 AM
Carl Gigun

I think he means a cap from the top of R2 to ground. I guess if you selected suitable charge/discharge time constants it would smooth the voltage change enough to prevent the earthquake? I haven't tried it yet, that's just a guess, but it seems like getting the rate of change down lower than any bass frequency should keep it from looking like a signal.  
-Carl Gigun

  Page 1 of 5 Next> Last Page>>