Joe Berry's Circlotron

Jump to Latest
#21
1648452762314.png
 
#24
jpatay said:
This topic is already there....
Click to expand...
Good !
So why do you bring it in here ?
This is about different circuit.

BTW, your amp is not without feedback - the output stage works with 100% local feedback, R7/R8 introduce gobs of local feedback/degeneration too. So you got tons of feedback, circuit merely omits the loop over the stages. If you want that kind of sound a single ECC88/6DJ8 does that job better than handful of JFETs...
 
Last edited:
  • Like
Reactions: userguide
#26
jpatay said:
It would be interesting to try your circuit without feedback.
Click to expand...
Hi jpatay,

My early MOSFET circlotron designs were mostly class A and did not use global negative feedback. They sounded quite good with the few pairs of speakers I had on hand for testing. Later on, as I gained experience with more difficult loudspeaker loads, I found that these amplifiers could lose some control when pushed, causing the sound to "fall apart" at high volumes.

Thinking about this problem led me to realize a limitation of the circlotron topology that I had not previously considered or seen discussed in the literature. Being essentially two single-ended amplifiers driving opposite sides of a load, this topology works fine as long as both sides are continuously active, i.e., operating class A. As soon as one side leaves class A and turns off, its side of the load is no longer under direct active control, making it susceptible to back-EMF and other load-dependent variables.

Any such influences from the load can only be "seen" by the still-active half of the output stage by "looking" through the load impedance itself. This severely limits the ability of the still-active half to maintain control of the full signal across the load. The only solutions for this that I can think of are either 1) to ensure that the amp operates pure class A under all intended load conditions, or 2) to provide some way to allow each half of the output stage to respond to conditions on both sides of the load.

In the real world, I can't ensure load conditions that will allow for full-time class A operation. So, all my later designs use some kind of feedback that gives each side of the output stage some "insight" into conditions on both sides of the load. In the design discussed here, this feedback is applied at the input stage, as global feedback is already needed there to define the amplifier's overall voltage gain.

Other circlotron designs exist (including mine) that use a local loop around the output stage alone -- or the internal equivalent, as provided by output devices having a "triode" characteristic -- but these probably deserve their own separate discussion.
 
  • Like
Reactions: jpatay and geraldfryjr
#29
jpatay said:
The amp sounds great, but I prefer the sound of amps without overall feedback.
Click to expand...
Understood, and such designs can work well in many situations. It wasn't until I had problems with difficult loads that I added any loop feedback at all.

Since then I've seen other circlotron designs that address the same problem without explaining it in detail. For example, GRAAF's description of their GM 20 amplifier on their old (and unsecure) website states that "A small amount of feedback (6dB) is used to allow the GM 20 to drive even difficult loudspeaker loads without problems."

From schematics of the GM 20 posted on DIYAudio here and here, you can see that its feedback loop includes the driver and output stages but (arguably) not the input stage, which should help preserve some of the "open loop" character of the sound.

Please stay safe -- I hope you can return to building amps soon!
 
  • Like
Reactions: jpatay and geraldfryjr
#31
As a mere "solder slinger" trying to follow in the footsteps of Sir Juma, I have a few doubts/questions.
I would also appreciate an "explanation" of the working of the circuit by the Master.

What are the supply rail max/min? And current requirement? I have a few good transformers and PSUs, hence the question.
(BTW, what is the output power of #13 circuit? Wanted to know if it can be made with enough output to drive a slot-loaded, multi-driver bass unit that supplements my full range speaker. In this context, will the circuit safely drive 16 Ohm load?)

I am assuming the circuit operates in class-AB (?), and that would mean heatsink requirements are more modest...?
What is the ideal--one device per heatsink, or both on the same; should the thermal coupling of the output devices be close?

What is the drive voltage requirement for full output?

Upon checking, I find that IRF9610, IRF9630 and IRFP054N are available from my sources. Is it possible to use them?
Any caveats while wiring up a P-2-P version?

With advance thanks for advice and guidance, and apologies for any "foolish" questions.
 
#32
juma said:
BF862 JFETS, being high-gm, are too twitchy and temperature sensitive, so in this circuit they cause DC offset on the speaker to wander around more than I like (and them being SMD doesn't help to thermo-couple them). It is just +/-30mV but when environmental temperature changes it needs readjusting.

So, for the peace of mind, I included the DC servo that solves the problem - now it moves between 2.3mV and 3.6mV no matter what the temp. conditions are.

DC servo makes a circuit more complex (OpAmp needs auxiliary +/-12V, 10mA PSU ). Sure it could be made discrete, without the OpAmp and without additional +/-12V PSU but parts number/complexity would be higher.

The circuit now looks like this:
Click to expand...
Nice work, Juma and Joe Berry. I love simple amps - could almost qualify as a “fistful of solder” type amp (parts fit in a fist). Would single package dual JFET like LSK389 reduce thermal drift? I have some on hand.
 
#35
minek123 said:
Anyone tried different output fets (IRFP044N) ?
Perhaps IRF260 would work? Have plenty of them..
Click to expand...
Any vertical enhancement mode MOSFET should work, possibly with adjustment (or removal) of the compensation caps. For reference, the original devices used were IRFP150s and they worked fine. FWIW, for my low voltage-design (nominal 22V DC rails), I'm using a logic-level MOSFET (RFP12N10L) to maximize VAS headroom. Just be sure that the device's maximum Vds rating is at least double the DC rail voltage.
 
Last edited:
#39
The Prof said:
.. I have a few doubts/questions....
Click to expand...

1. What are the supply rail max/min?
Depends on maximal Vce/Vds for devices used in the circuit (study the datasheets).

2. And current requirement?
Strictly Ohm's Law compliant -> I=V/R

3. what is the output power of #13 circuit?
About 40W @ 4R load

4. will the circuit safely drive 16 Ohm load?
Yes. More ohms -> less amperes. High time to understand the Ohms Law

5. I am assuming the circuit operates in class-AB (?), and that would mean heatsink requirements are more modest...?
Can be called shallow cl. A or deep cl. AB. Less amperes -> less heat (add Joule's Law to the to-learn list)

6. What is the ideal--one device per heatsink, or both on the same; should the thermal coupling of the output devices be close?
Output MOSFETs on one heatsink and Q7 between them.

7. What is the drive voltage requirement for full output?
Depends on chosen gain. For circuit in post #13 it's 1V

8. Upon checking, I find that IRF9610, IRF9630 and IRFP054N are available from my sources. Is it possible to use them?
Yes, I'd choose IRFP054N

9. Any caveats while wiring up a P-2-P version?
Just follow the schematic correctly

Dear Prof, AFAIC you used up all you questions. Holding hands across the street is so comfortable but doing it on your own brings more joy. Good luck, have fun :cheers:
 
#40
@ Juma
You are the quintessential Guru ... not at all interested in imposing your take...leaving the 'disciple' to the task of study, understanding and discovery, with some light shone on the path!! I am wordless in my thanks...

Can I substitute IRF9620 with 9630?
Thanks and warm regards.
 
You must log in or register to reply here.
Top Bottom