djk said:
The schematic in this post looks to me to be essentially a Sumo Nine topology done with tubes instead of opamps, and a pair of large vertical MOSFETs in place of two banks of bipolar output devices.
Bias for the output stage here appears to be controlled by a single adjustable constant current source. Assuming that the CCS is set up to thermally track the output MOSFETs, this scheme should keep their bias stable enough to allow for safe class AB operation.
Out of curiosity, I modeled a version the circuit I mentioned above in Electronics Workbench and it seems to work fine. I've attached a screen shot of the model for anyone interested. I used the basic Sumo Nine circuit, replaced the output stage with a pair of IRF250s, and replaced the Sumo bias circuit with a CCS model. This CCS draws about 3mA from each opamp with the output stage idling at 0.9A per side. The model CCS is not temp compensated, but there are several ways this could be done in practice.
For any reasonable load (>1 ohm), the output stage acts as a current bootstrap over the opamps so their output current never exceeds twice their 3mA idle current, allowing them to work class A at all times. A side benefit of this is that the output stage Vgs never exceeds twice its idle value, so there is a degree of protection for the output stage built into the circuit. This happy scenario may not hold up against a dead short at the output, so in practice I'd add a zener Vgs clamp for each output MOSFET. And although the model needed none, I'd expect to see the need for some HF compensation as shown in the original Sumo Nine schematic to prevent HF instability.
The objective performance of the circuit looks great in the model, and from past experience, I think most of this would translate to reality. However, I did have to set the model LF353's operating voltage 2V higher than in the Sumo to get 60W output without clipping. So, anyone modifying an existing Sumo Nine may see a slight decrease in max output voltage swing, this being due to the 4V turn-on voltage of the output MOSFETs vs. the .6-1.3V of the original bipolars.
For any reasonable load (>1 ohm), the output stage acts as a current bootstrap over the opamps so their output current never exceeds twice their 3mA idle current, allowing them to work class A at all times. A side benefit of this is that the output stage Vgs never exceeds twice its idle value, so there is a degree of protection for the output stage built into the circuit. This happy scenario may not hold up against a dead short at the output, so in practice I'd add a zener Vgs clamp for each output MOSFET. And although the model needed none, I'd expect to see the need for some HF compensation as shown in the original Sumo Nine schematic to prevent HF instability.
The objective performance of the circuit looks great in the model, and from past experience, I think most of this would translate to reality. However, I did have to set the model LF353's operating voltage 2V higher than in the Sumo to get 60W output without clipping. So, anyone modifying an existing Sumo Nine may see a slight decrease in max output voltage swing, this being due to the 4V turn-on voltage of the output MOSFETs vs. the .6-1.3V of the original bipolars.
Attachments
FWIW, scratch-builders can increase the power output available from this topology by specifying higher rail voltages and replacing the chip opamp front end with something similar that can handle the higher voltage.
For example, the attached schematic shows a 200W model in which the chip opamps are replaced by discrete FET/MOSFET circuits, the output rails are 72V each, and the front end supply is +/-40V. The other circuit values were chosen rather casually, but even so, the model works well and predicts excellent measured performance.
For example, the attached schematic shows a 200W model in which the chip opamps are replaced by discrete FET/MOSFET circuits, the output rails are 72V each, and the front end supply is +/-40V. The other circuit values were chosen rather casually, but even so, the model works well and predicts excellent measured performance.
Attachments
djk said:
I agree that it would if the bias CCS had a zero temperature coefficient. My thought is to make the CCS tempco a negative complement of the OS tempco, probably using a small HEXFET mounted near the OS heatsink as the CCS voltage reference.
I have to say that I haven't yet built this particular circuit, and until I do I'll keep an open mind. However, the amp I listen to most often these days uses a hot-running HEXFET circlotron OS with IRFD110 bias. The amp has been on more or less continuously for over three years (in and out of standby mode), and in that context at least, bias stability is a non-issue.
I'll take it if the freight is not too much.
Check freight to 71854.
thanks, sdman
sdman@cableone.net
Check freight to 71854.
thanks, sdman
sdman@cableone.net
The US Post Office has a flate-rate Priority Mail box for $7 or so that will hold a much bigger transformer, just pack it in crumpled newspaper.
The Sumo transformer is also used in the Andromeda. It is 1.4KVA at 50% duty cycle with 30-30-30-30 and 20-0-20(at low VA).
Bongiorno used the same transformer, heatsink, and chassis for the Andromeda as well. The Andromeda was a 250W class AB amplifier vs the 9 being class A at 60W~70W(9 or 9+).
The 9 is a sliding bias push-pull class A and so is quite efficient as far as class A amplifiers go.
The Sumo transformer is also used in the Andromeda. It is 1.4KVA at 50% duty cycle with 30-30-30-30 and 20-0-20(at low VA).
Bongiorno used the same transformer, heatsink, and chassis for the Andromeda as well. The Andromeda was a 250W class AB amplifier vs the 9 being class A at 60W~70W(9 or 9+).
The 9 is a sliding bias push-pull class A and so is quite efficient as far as class A amplifiers go.
It doesn't weigh that much, 12~15 lbs or so. The USPO will sent any weight you can fit in that box for the flat rate.
If trying to build one with an opamp, remember the LF353 was selected to work on ±20V. Use a ±24V rated OPA2604 or select an AD823 for ±20V operation. Either will sound vastly better than an LF353.
If trying to build one with an opamp, remember the LF353 was selected to work on ±20V. Use a ±24V rated OPA2604 or select an AD823 for ±20V operation. Either will sound vastly better than an LF353.
new amp is a circlotron but different to Sumo
Hallo Joe,
regarding post no. 61: the picture is coming from a test article in german hi-fi magazin "stereoplay". At the moment (and I guess for a limited time only) you can download the test for free here:
http://www.stereoplay.de/sixcms/media.php/188/stp0106Thorens.pdf
If this link does not work go to www.stereoplay.de , click onto the eye catching picture of the new Thorens TEM 3200 amp and then go to the end of the preview text to the download button. The test is in german, of course, but it has some nice pictures and maybe google translator could help a little.
From my point of view this new amp is different to the Sumo in some major respects:
- The Sumo Amp will never work with Mosfets, because the Sumo ist controlling the base current of output BJTs only (see Bongiorno patent). The necessary gate-source voltage for biasing the output Mosfets is not present.
- The Sumo uses 2 different potentiometers, connected to a voltage potential, for setting the quiescent current of the output stage. The amp from Mr. Bloehbaum (see test) uses 1 variabel current source only for setting the quiescent current of the 2 floating output stages. That is a major step forward in biasing a DC-coupled circlotron.
- The Sumo amp is limited from about 40 (nine) to about 70 (nine+) Watts rms (4 Ohm) only due to its inherent structure. The new Thorens Amp delivers 400 Watts rms at 4 Ohm and is absolutely stable at 2 Ohm loads. This power is never possible with the Sumo amp structure.
- The new amp is fully DC-coupled and uses 6 tubes in the voltage amplifier and driver stage. That would never be possible with the Sumo structure because the termal drift of the tube working points would cause unwanted current flows in the output stage in terms of many Amperes.
- The Sumo amp has a sudden switch point from class-A to class-AB which definately causes audible distortions. The very good near constant slope of the harmonic spectral curves of the new Thorens amp would never be possible with the Sumo amp structure.
arcolette
Hallo Joe,
regarding post no. 61: the picture is coming from a test article in german hi-fi magazin "stereoplay". At the moment (and I guess for a limited time only) you can download the test for free here:
http://www.stereoplay.de/sixcms/media.php/188/stp0106Thorens.pdf
If this link does not work go to www.stereoplay.de , click onto the eye catching picture of the new Thorens TEM 3200 amp and then go to the end of the preview text to the download button. The test is in german, of course, but it has some nice pictures and maybe google translator could help a little.
From my point of view this new amp is different to the Sumo in some major respects:
- The Sumo Amp will never work with Mosfets, because the Sumo ist controlling the base current of output BJTs only (see Bongiorno patent). The necessary gate-source voltage for biasing the output Mosfets is not present.
- The Sumo uses 2 different potentiometers, connected to a voltage potential, for setting the quiescent current of the output stage. The amp from Mr. Bloehbaum (see test) uses 1 variabel current source only for setting the quiescent current of the 2 floating output stages. That is a major step forward in biasing a DC-coupled circlotron.
- The Sumo amp is limited from about 40 (nine) to about 70 (nine+) Watts rms (4 Ohm) only due to its inherent structure. The new Thorens Amp delivers 400 Watts rms at 4 Ohm and is absolutely stable at 2 Ohm loads. This power is never possible with the Sumo amp structure.
- The new amp is fully DC-coupled and uses 6 tubes in the voltage amplifier and driver stage. That would never be possible with the Sumo structure because the termal drift of the tube working points would cause unwanted current flows in the output stage in terms of many Amperes.
- The Sumo amp has a sudden switch point from class-A to class-AB which definately causes audible distortions. The very good near constant slope of the harmonic spectral curves of the new Thorens amp would never be possible with the Sumo amp structure.
arcolette
Re: new amp is a circlotron but different to Sumo
Arcolette,
Thanks for the link and for your comments. I certainly agree that the Thorens amp is not identical to the Sumo. I continue to see it as based on the same basic topology, but I make this comparison only as an aid to understanding, not to detract from the later design in any way or to suggest that it trades unfairly on the Bongiorno patent (which has expired in any case). If you compare the schematics of these amps to most of what's out there, you may see why I think of them as relatively similar.
As mentioned earlier, I believe that the Thorens is using a pair of vertical MOSFETs biased by a temperature-compensated current source. I'm sure it sounds great with a tube voltage gain section, but for what it's worth, you could get the same power output using a discrete solid-state front end (see example at post #64).
P.S. - Is an English translation available for the review article you linked to above?
Arcolette said:
Arcolette,
Thanks for the link and for your comments. I certainly agree that the Thorens amp is not identical to the Sumo. I continue to see it as based on the same basic topology, but I make this comparison only as an aid to understanding, not to detract from the later design in any way or to suggest that it trades unfairly on the Bongiorno patent (which has expired in any case). If you compare the schematics of these amps to most of what's out there, you may see why I think of them as relatively similar.
As mentioned earlier, I believe that the Thorens is using a pair of vertical MOSFETs biased by a temperature-compensated current source. I'm sure it sounds great with a tube voltage gain section, but for what it's worth, you could get the same power output using a discrete solid-state front end (see example at post #64).
P.S. - Is an English translation available for the review article you linked to above?
"The Sumo amp is limited from about 40 (nine) to about 70 (nine+) Watts rms (4 Ohm) only due to its inherent structure. "
The 9+ does 120W at 4 ohms.
" This power is never possible with the Sumo amp structure."
The base current set can be increased to put out any amount of power that you want for low impedances, there is enough transformer, heatsink, and outputs to drive down to 1 ohm if you wish.
"The Sumo amp has a sudden switch point from class-A to class-AB which definately causes audible distortions. "
Yours may be adjusted wrong, I don't have that problem with mine.
The 9+ does 120W at 4 ohms.
" This power is never possible with the Sumo amp structure."
The base current set can be increased to put out any amount of power that you want for low impedances, there is enough transformer, heatsink, and outputs to drive down to 1 ohm if you wish.
"The Sumo amp has a sudden switch point from class-A to class-AB which definately causes audible distortions. "
Yours may be adjusted wrong, I don't have that problem with mine.
Hallo djk,
“The 9+ does 120W at 4 ohms.”
I talked about rms Watts, you too? My amp delivers 60W rms at 4 ohms and 95W rms at 8 ohms only. The available output power strongly depends on the current gain of the output transistors and on the transfer characteristic of these transistors and verys from amp to amp. The higher the output current the lower the current gain (Ic/Ib) of these transistors as usual with BJTs. But the base current is fixed by the structure of this amp and can increase linearly with the output voltage only – at first hand looking. That is much too low for higher output power or in better terms higher output current (because the current gain drops rapidly with collector current). For this reason the Sumo has 2 additional diodes and 2 current limiting resistors connected from the output terminals to ground. This circuit opens a current path whenever the output voltage reaches the forward voltage drop of these diodes (+/- 0.7 V to ground). This additional current delivers the necessary base current for higher output loads. But even this is only good for about 40W rms at 4 ohms (Sumo 9). Bongiorno discovered that and added the very clever designed additional current circuit including 2 pnp small signal transistors. These transistors add (switch!) a third current path at an output power of about 10W rms at 4 ohms. How much power is possible at the end depends from your used power BJTs. But with 5 output transistors and all the other used circuits it is never possible to drive a 1 ohm load sufficiently due to its inherent current limiting functionality.
“ The base current set can be increased to put out any amount of power that you want for low impedances, there is enough transformer, heatsink, and outputs to drive down to 1 ohm if you wish.”
No, I fully disagree. See my notes above. The circuit works inherent current limiting. Increasing the output current means adding a lot(!) of additional power output transistors and additional circuits for delivering the necessary base current. One has to increase the quiescent current of the output transistors to at least 1 A per transistor. That would cause way too much heat and the driving circuit is not able to deliver the necessary base current. I made some simulations and they delivered exactly the same results. Even by using most modern BJTs with a fairly good Ic/Ib ratio (MJL3281A as an example) does not solve this structural problem.
Regarding the switching effects you wrote: “ Yours may be adjusted wrong, I don't have that problem with mine.”
No, the switching problem has nothing to do with adjustments. There are 2 base current paths for Sumo 9 and 3 for Sumo 9+. Only one is stable, the other ones are switched suddenly depending on output voltage and current (see my comments above). The second current path is switched at about 0.7 Volts refering to ground or 1.4 Volts across the load (speaker) with the mentioned diode circuit. The third circuit incorporating the additional pnp transistors in Sumo 9+ switch at about 10 Watts rms at 4 Ohms (depending on current gain ratio of output transistors). I do not like any switching which depends on output power. That is never a good idea for a no-compromise good sounding amplifier because it definitely has bad effects onto the output signal. Please take your time and make a simulation, you will clearly see this effect. Another drawback is that the performance of this circuit strongly depends on the near perfect matching of 10(!) output power transistors. That is a very high effort for an amplifier which has to make large compromises (switching!) to reach a fairly amount of output power.
“The 9+ does 120W at 4 ohms.”
I talked about rms Watts, you too? My amp delivers 60W rms at 4 ohms and 95W rms at 8 ohms only. The available output power strongly depends on the current gain of the output transistors and on the transfer characteristic of these transistors and verys from amp to amp. The higher the output current the lower the current gain (Ic/Ib) of these transistors as usual with BJTs. But the base current is fixed by the structure of this amp and can increase linearly with the output voltage only – at first hand looking. That is much too low for higher output power or in better terms higher output current (because the current gain drops rapidly with collector current). For this reason the Sumo has 2 additional diodes and 2 current limiting resistors connected from the output terminals to ground. This circuit opens a current path whenever the output voltage reaches the forward voltage drop of these diodes (+/- 0.7 V to ground). This additional current delivers the necessary base current for higher output loads. But even this is only good for about 40W rms at 4 ohms (Sumo 9). Bongiorno discovered that and added the very clever designed additional current circuit including 2 pnp small signal transistors. These transistors add (switch!) a third current path at an output power of about 10W rms at 4 ohms. How much power is possible at the end depends from your used power BJTs. But with 5 output transistors and all the other used circuits it is never possible to drive a 1 ohm load sufficiently due to its inherent current limiting functionality.
“ The base current set can be increased to put out any amount of power that you want for low impedances, there is enough transformer, heatsink, and outputs to drive down to 1 ohm if you wish.”
No, I fully disagree. See my notes above. The circuit works inherent current limiting. Increasing the output current means adding a lot(!) of additional power output transistors and additional circuits for delivering the necessary base current. One has to increase the quiescent current of the output transistors to at least 1 A per transistor. That would cause way too much heat and the driving circuit is not able to deliver the necessary base current. I made some simulations and they delivered exactly the same results. Even by using most modern BJTs with a fairly good Ic/Ib ratio (MJL3281A as an example) does not solve this structural problem.
Regarding the switching effects you wrote: “ Yours may be adjusted wrong, I don't have that problem with mine.”
No, the switching problem has nothing to do with adjustments. There are 2 base current paths for Sumo 9 and 3 for Sumo 9+. Only one is stable, the other ones are switched suddenly depending on output voltage and current (see my comments above). The second current path is switched at about 0.7 Volts refering to ground or 1.4 Volts across the load (speaker) with the mentioned diode circuit. The third circuit incorporating the additional pnp transistors in Sumo 9+ switch at about 10 Watts rms at 4 Ohms (depending on current gain ratio of output transistors). I do not like any switching which depends on output power. That is never a good idea for a no-compromise good sounding amplifier because it definitely has bad effects onto the output signal. Please take your time and make a simulation, you will clearly see this effect. Another drawback is that the performance of this circuit strongly depends on the near perfect matching of 10(!) output power transistors. That is a very high effort for an amplifier which has to make large compromises (switching!) to reach a fairly amount of output power.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.