Well, this design has something to make pretty much everybody angry. Op-amps, SE transformer, mosfets. 😀
After my experiments here I have a very low distortion op-amp driving an EL34 that is capable of huge swings at very low distortion on the bench.
My next experiment on this bench setup was to modify a bias servo that I built to keep the EL34 plate voltage right on spec and use it to bias a mosfet output stage (to see if it would work well for converting my Unity-Coupled amp to mosfet output devices).
After doing that experiment, it would be trivial to throw an SE transformer on the mosfet and build an SE source follower bench amp. Of course, the first stage could easily be replaced with any tube stage you might wish, leaving only solid-state followers and CCS loads in the design if one does not like amplification done by solid-state components (although the ADA4700 has beautiful distortion performance in my opinion, low distortion and it is low-order, second harmonic dominant).
First question: I've never built an SE amp before. I'd like to get a decent transformer and I'll be able to swing a good bit more voltage than a 300B (since the mosfet saturates at a lower voltage). My calculations show 17+ Watts from a 5k transformer (minus transformer losses). I don't want to under-size the transformer. Any recommendations? I was thinking of going with an Electra-Print 5k:8 20W. I'm okay with not going the cheap route here.
After my experiments here I have a very low distortion op-amp driving an EL34 that is capable of huge swings at very low distortion on the bench.
My next experiment on this bench setup was to modify a bias servo that I built to keep the EL34 plate voltage right on spec and use it to bias a mosfet output stage (to see if it would work well for converting my Unity-Coupled amp to mosfet output devices).
After doing that experiment, it would be trivial to throw an SE transformer on the mosfet and build an SE source follower bench amp. Of course, the first stage could easily be replaced with any tube stage you might wish, leaving only solid-state followers and CCS loads in the design if one does not like amplification done by solid-state components (although the ADA4700 has beautiful distortion performance in my opinion, low distortion and it is low-order, second harmonic dominant).
First question: I've never built an SE amp before. I'd like to get a decent transformer and I'll be able to swing a good bit more voltage than a 300B (since the mosfet saturates at a lower voltage). My calculations show 17+ Watts from a 5k transformer (minus transformer losses). I don't want to under-size the transformer. Any recommendations? I was thinking of going with an Electra-Print 5k:8 20W. I'm okay with not going the cheap route here.
Attachments
why you don´t drive pp opt-no gap directly from mu out /ccs/
Both upper fets are not needed.
Rather medium mu triode than opamp.
Both upper fets are not needed.
Rather medium mu triode than opamp.
Last edited:
https://www.fairchildsemi.com/datasheets/FQ/FQP3P50.pdf is ok.
but in this schade, low cap is nice to have.
VP0550N3-G MICROCHIP (SUPERTEX) - Tranzistor: P-MOSFET | TME Slovakia s.r.o. - Elektronické sú?iastky
but in this schade, low cap is nice to have.
VP0550N3-G MICROCHIP (SUPERTEX) - Tranzistor: P-MOSFET | TME Slovakia s.r.o. - Elektronické sú?iastky
Mostly because I wanted to see how good an amp I could make with a typical gapped SE transformer. I want to see how well I can do it, compare results with other amps. It is a game I am playing with myself. I do not think it is the best approach, but I have no doubt that it will produce a very listenable amp with low distortion. The other reason for not taking the output from the EL34 is that I am only running it at 10mA. The Fet will be running at an operating point would fry the EL34 and the EL34 would not be able to drive a 5k load to very high power without biasing it really hot, which my bench setup is not prepared for.
The circuit with the two upper fets is what I have on the bench at this moment. I was going to do an experiment to evaluate the two fets vs. taking feedback from mu output to make sure that performance still stays as good. Haven't done that yet, but I don't see any reason at the moment that it wouldn't be a great thing to simplify it and eliminate the other fet follower. I have experienced a bit of peaking in the frequency response with the higher-capacitance (and higher voltage) fets that I just installed and taking that second fet out might help calm that down a bit.
I realized that many would not like the opamp input stage, but it is what I have on the bench at the moment and it does have very good (and very low order) distortion performance. I probably will eventually build this into an amp and it doesn't have to be in the final version, just this breadboard amp.
Last edited:
I think the VP0550N3 would be fried with the supply voltages and feedback currents that I am using on the bench amp but it is a possible suggestion for a re-worked circuit. I don't think the opamp will have any issue with the Fairchild part at least for this initial experiment.
ok in that case
this pfet schade is best idea out there (equivalent of driver transformer version)
this pfet schade is best idea out there (equivalent of driver transformer version)
It works extremely well for driving the low input impedance of the parallel applied feedback. Did you ever see the amp I built where I used this in the output stage? Link here.
I'm also really happy about how well it has worked out with CCS loaded pentodes as well. In the past I have had a hard time finding triodes that can do really wide swings with low distortion. Now I can use an EL84 or EL34 for pretty much anything and they are cheap and in production.
But as for my question above, I was thinking of getting the 20W Electra-Print gapped SE transformers. Is that a good choice for something that will be able to drive the transformer to 17+ Watts or should I look at something bigger? Edcor claims rating to 25W, would that be a better choice?
I just have no practical experience with gapped SE transformers. This is probably my one and only SE amp I will build and I don't want to buy the wrong transformers. I have heard good things about Electra-Print but the mosfet will also be able to push them harder than a 300B. I will probably just call and explain my application before purchasing but I wanted to see if there were other known good 5k transformers for this approximate power level.
I just have no practical experience with gapped SE transformers. This is probably my one and only SE amp I will build and I don't want to buy the wrong transformers. I have heard good things about Electra-Print but the mosfet will also be able to push them harder than a 300B. I will probably just call and explain my application before purchasing but I wanted to see if there were other known good 5k transformers for this approximate power level.
Alright, so no suggestions for an Electra-Print alternative for a transformer that might get pushed a bit beyond typical 300B voltage swing levels?
I'm getting everything gathered to start testing my heatsink/mosfet/bias servo to make sure I can manage everything thermally in the output stage. This will be my first attempt at 35W+ dissipation in a solid-state device so should be fun. I got a matched pair of heatsinks I salvaged from some motor drivers so I hope they are big enough.
Then I'll probably call up Jack at Electra-Print and either purchase the 20W 5k transformer that people use for 300Bs or have him build something a bit bigger if that makes more sense.
I'm getting everything gathered to start testing my heatsink/mosfet/bias servo to make sure I can manage everything thermally in the output stage. This will be my first attempt at 35W+ dissipation in a solid-state device so should be fun. I got a matched pair of heatsinks I salvaged from some motor drivers so I hope they are big enough.
Then I'll probably call up Jack at Electra-Print and either purchase the 20W 5k transformer that people use for 300Bs or have him build something a bit bigger if that makes more sense.
Last edited:
I bought some fets to try out with this feedback scheme on my last Mouser order. I know that the increased Cin will present some problems but I will try them out to see how well they perform.
Well, this amp is coming to life. I just got some 5k:8 Edcor output transformers in. They are supposedly rated for 25W. We'll see if they can handle 20.
I measured DCRs today. Primary DCR was 86 Ohms. Secondary was 0.39 Ohm.
We'll see how low the Zout of this amp ends up but I know that it absolutely won't be less than 0.53 Ohm.
I measured DCRs today. Primary DCR was 86 Ohms. Secondary was 0.39 Ohm.
We'll see how low the Zout of this amp ends up but I know that it absolutely won't be less than 0.53 Ohm.
I didn't see any of the early postings in this thread until today.
I have a pair of the Edcor CXSE25-5K 8 ohm SE transformers. I know they will handle a KT88 pushed to 16 watts in UL on 500 volts down to about 30 Hz with some cathode feedback to lower the output impedance and keep my 15 inch woofers in line.
The big Hammond 1628SEA's have a little more room at the low frequencies, but the high frequency response just sucks.
I have also breadboarded a similar "Schade" feedback scheme using a mosfet. It worked quite well until the pot I was using to set the feedback ratio caught on fire. I'll get back to those experiments some day.
The Zout can be less than the DCR if you apply feedback from the OPT secondary. ( just something else to rattle the SE purists)
I have a pair of the Edcor CXSE25-5K 8 ohm SE transformers. I know they will handle a KT88 pushed to 16 watts in UL on 500 volts down to about 30 Hz with some cathode feedback to lower the output impedance and keep my 15 inch woofers in line.
The big Hammond 1628SEA's have a little more room at the low frequencies, but the high frequency response just sucks.
I have also breadboarded a similar "Schade" feedback scheme using a mosfet. It worked quite well until the pot I was using to set the feedback ratio caught on fire. I'll get back to those experiments some day.
The Zout can be less than the DCR if you apply feedback from the OPT secondary. ( just something else to rattle the SE purists)
IMHO, the schematic looks unnecessarily complicated, why a opamp, a tube, fet's....Different topologies, kind of circuits, distortion,... Simpler circuits works pretty fine.
I had read some stuff on your website about comparative measurements of output transformers that helped me settle on the Edcor transformers. Thanks for putting that stuff out there.
Ha! I smoked my 100k feedback resistor testing this circuit. I just overlooked the fact that the dissipation in the resistor is very high until the plate voltage comes down as the EL34 cathode gets to temperature. I didn't have problems until I went to the 1000V supply. Now I am using a 25W thick film resistor on a heatsink. The smoke stays in that one.
I think I explained that above. I had been experimenting on CCS loaded EL34s with plate-grid feedback and had made a low distortion op-amp driver (for the purpose of making distortion measurements) and then just thought to myself, "It would be super easy to tack an output stage onto this and see how an SE amp could be made with a super-low transformer driving impedance and minimal distortion." Plus, I wanted to see how my bias servo board would do at biasing a mosfet output stage. There are lots of little things I'm getting out of this weird experiment.
And in the end I think this will produce at least an order of magnitude less distortion than more conventional SE circuits, maybe two orders. Plus, I don't really think it is all that complicated. So that's why the extra complication. 😉
I'm not saying people should build this but I thought it would be an interesting and easy experiment since it was practically already finished. It could, however, be easily modified to have a tube front end and a tube follower output stage.
Well, I powered it up last night and had a listen. The only source I had available in the garage was my cell phone which doesn't have enough output to get this amp above 3W output or so. Amp sensitivity isn't high enough for direct use with cell phones. I hooked it up to a JBL JRX-215, which is a PA speaker but is still really quite efficient and has a 15 inch woofer and a tweeter, rated for 250W continuous or something monstrous.
I have to say that I didn't feel that there was anything lacking in the lows or the highs. With respect to frequency response, it sounds very similar to my Unity-Coupled amp at a similar volume level.
Subjectively, it sounds very nice, balanced, clear, and detailed. I listened to some bass-heavy electronic music and it showed no signs of breaking up or having difficulty in any way. It was able to drive the speaker to a comfortable listening level with only a few Watts output. This is probably where I would set the volume if I were listening to the amp while working in the workshop. I am thinking I will probably design some boards for the driver and build this into a chassis and will probably pair it with some old '70s Sansui speakers that I have and maybe use it as a workshop amp.
I'll have to think about what I want to do with the input stage, whether I want to keep the ADA4700 or put a tube there.
One thing I did notice is that the output mosfet was getting the heatsink up to ~65 deg C and the package was getting up around 100 deg C. I'll probably need to get a bigger one if I want to mount it external to a case.
Measurements to follow in the next few weeks...
I have to say that I didn't feel that there was anything lacking in the lows or the highs. With respect to frequency response, it sounds very similar to my Unity-Coupled amp at a similar volume level.
Subjectively, it sounds very nice, balanced, clear, and detailed. I listened to some bass-heavy electronic music and it showed no signs of breaking up or having difficulty in any way. It was able to drive the speaker to a comfortable listening level with only a few Watts output. This is probably where I would set the volume if I were listening to the amp while working in the workshop. I am thinking I will probably design some boards for the driver and build this into a chassis and will probably pair it with some old '70s Sansui speakers that I have and maybe use it as a workshop amp.
I'll have to think about what I want to do with the input stage, whether I want to keep the ADA4700 or put a tube there.
One thing I did notice is that the output mosfet was getting the heatsink up to ~65 deg C and the package was getting up around 100 deg C. I'll probably need to get a bigger one if I want to mount it external to a case.
Measurements to follow in the next few weeks...
Certainly, I just want to get it in the house and take detailed measurements that are under comparable conditions to the measurements I have been taking before I change anything there. I don't want to change several things at once.
- Status
- Not open for further replies.