Respect!
Suzan, is this marble on which the EI transformers are bolted on?
https://www.diyaudio.com/community/...und-torroid-output-transrormer-1-jpg.1175987/
George
Suzan, is this marble on which the EI transformers are bolted on?
https://www.diyaudio.com/community/...und-torroid-output-transrormer-1-jpg.1175987/
George
Hi George,
The EI transformer amps and their power-supplies are made with steel wall-boxes, Schroff, etc. There is no reason they couldn't be made "breadboard" style on marble or other stone slabs.
The toroid bass-amp is using a cover from a 3U 19" rack case, which I will need back at some point for my next line-driver, "field expedient" is I believe the term 🙂
The EI transformer amps and their power-supplies are made with steel wall-boxes, Schroff, etc. There is no reason they couldn't be made "breadboard" style on marble or other stone slabs.
The toroid bass-amp is using a cover from a 3U 19" rack case, which I will need back at some point for my next line-driver, "field expedient" is I believe the term 🙂
I will have to build one eventually, so I will be able to experience it myself. 🙂
Would IRFP048NPbF work? This is what's being sold with the PASS DIY F6 kits. Although Pd is only 140W vs 180W...
How critical is it to match the MOSFETs. Could you add a "balance pot" from the bias supply to the two transformer inputs?
Most N MOSFETs will work, just some better than others... BUT they must have enough voltage headroom, as their Sources swing negative as well as positive... so this has to be double the power supply voltage plus a good safety margin.
Sadly the IRFP048NPbF part is specified with a Vdss at 55V which is too low.
The IRF150 with a Vdss of 100V is the minimum I would recommend, these were the type I originally used in my first versions.
I look carefully at the "Drain to Source Volts" graphs - one is looking for as flat as possible in the bottom 1 to 10A range lines. The "knee" in the left-hand side from zero to where it goes flat is the headroom voltage one is looking for on top of the normal power usage peaks on the primary side waveform.
My main stereo monoblock amps have about 38V to 40V DC supplies; so the maximum possible swing is c, 80V, and then a 20V margin would give 100V minimum.
Higher than 100V is fine, as one is looking at the overall parameters, up to c. 500V; The STW34NB20 is 200V, the IXTH20N50D is 500V.
The IRFP250MPBF looks interesting, definitely worth a try.
As to matching, some MOSFETs are excellent straight out of the tube, I found the STW34NB20 to be excellent in that regard. Others like the IXTH20N50D depletion MOSFETs are all over the place and one needs lots of parts to sort to get paired matches.
Up to 10mA preferred (20mA okay) difference in biasing at c. 750mA per MOSFET is fine with EI-120 laminations, which have a small nominal air-gap by the default of being E and I pieces stacked together. For a toroid core one would want as good as possible, which is why using mains types with dual 120V windings as the primaries are a problem since they will be of different lengths and thus resistance.
I did try making it adjustable between pairs, but it was adding pain and I wasn't convinced it was not causing more issues than it solved.
Looking at the IXTQ60 again.
What voltage is needed to drive the PA to full output? Could its input transformer be driven by a tube or SS preamp that has a cathodyne output (triode or LND150) in place of the line amplifier section?
What voltage is needed to drive the PA to full output? Could its input transformer be driven by a tube or SS preamp that has a cathodyne output (triode or LND150) in place of the line amplifier section?
Depends on the configuration, transformer specifications, and supply-voltage.
Yes, one can use a lower output impedance circuit to drive the power stage - a headphone output for example, or a follower. The input transformer as specified isn't designed for biasing though one could of course make something that is.
The line-driver output-transformer and the power-stage input-transformer are designed to operate as a pair - it is in effect a single transformer split into two parts. This minimizes the length of the step-up secondary-windings and the interwinding-capacitance, etc. which would otherwise be much greater if it was all on the same bobbin-core. I can then also optimize the core materials, i.e. M6 for the first and MuMetal for the second, which would be prohibitively more expensive if everything was on a single EI-96 core.
Depending on the output impedance you may need to adjust the power-stage's Rterm value to get the best performance, otherwise if mismatched you will get a roll-off or peak at the high-frequency end of things.
As an aside, our Oppo BD player has 2.1V ac max output, which is 2.95V peak-to-peak referenced to ground on the phono outs, but also has balanced XLR to the same 2.95V peak level which doubles the effective output signal.
This is basically what I was thinking.
I am interested in using this for guitar amplification, so bandwidth and distortion are less of a concern.
I think I read somewhere previously in this thread that the Line Amplifier section had an over gain of 4x? If that's true, then about 10v is needed at the input of the PA's input transformer to drive the PA to full output?
I am interested in using this for guitar amplification, so bandwidth and distortion are less of a concern.
I think I read somewhere previously in this thread that the Line Amplifier section had an over gain of 4x? If that's true, then about 10v is needed at the input of the PA's input transformer to drive the PA to full output?
Hi @stephenl5 ... Thanks for the circuit.
You have a phase-splitter driving a centre-taped transformer, which is itself a phase-splitter.
This isn't going to work as you might think, as the reflected impedance the tube sees is 600 ohms (depending on Rterm).
As it's a transformer, you only need one isolating capacitor - the two in series halves the capacitance so doubles the LF cutoff.
TubeCAD circuit SE Phase-Splitter
From here: https://www.tubecad.com/2018/07/blog0433.htm "Extra Super Simple Tube Circuits" about 3/4 the way down the page.
Did you consider having the tube phase-splitter directly driving the MOSFET inputs?
You have a phase-splitter driving a centre-taped transformer, which is itself a phase-splitter.
This isn't going to work as you might think, as the reflected impedance the tube sees is 600 ohms (depending on Rterm).
As it's a transformer, you only need one isolating capacitor - the two in series halves the capacitance so doubles the LF cutoff.
TubeCAD circuit SE Phase-Splitter
From here: https://www.tubecad.com/2018/07/blog0433.htm "Extra Super Simple Tube Circuits" about 3/4 the way down the page.
Did you consider having the tube phase-splitter directly driving the MOSFET inputs?
Thanks Susan, I'm a long time Broskie fan 🙂.
I did consider going directly to the MOSFETs, but the input transformer makes biasing easy and also I'd like to have the 300v preamp in a separate enclosure from the 35v power amp.
I did consider going directly to the MOSFETs, but the input transformer makes biasing easy and also I'd like to have the 300v preamp in a separate enclosure from the 35v power amp.
Yes, keeping the B+ safe is a good consideration 🙂
Remember the input impedance of the power-stage is c. 600 ohms, so if you put in a 600 ohm resistor into the schematic instead of the XLR... ???
Remember the input impedance of the power-stage is c. 600 ohms, so if you put in a 600 ohm resistor into the schematic instead of the XLR... ???
Interesting to see it as a voltage follower implementation. Most have the source or emitter grounded with power on the center tap.
As a follower, a concertina setup could be used to drive both sides. As shown in the post above but without transformer both outputs to the gates with blocking cap.
As a follower, a concertina setup could be used to drive both sides. As shown in the post above but without transformer both outputs to the gates with blocking cap.
LTspice simulation, note adjusted values and using DN2540:
I can't say if this will work, or how well.
Note that R8 at 600 (ohms) represents the power-stage input-impedance as seen by the line-driver.
I can't say if this will work, or how well.
Note that R8 at 600 (ohms) represents the power-stage input-impedance as seen by the line-driver.
Do you think I could add in a balance pot like this to compensate for device and xformer wire length differences? You mentioned that maybe you had some issues with trying methods to force balance? Any suggestions on R value?
I just put them into the LTspice standard.mos file, at the top. Here's mine to copy from - first five .MODEL entries.
(Depending on the version of LTspice, you may have a different set of files.)