Slewmaster - CFA vs. VFA "Rumble"

[Krisfr]

OS

Will you put up your best OPS with the FET outputs and let someone build it.

Do not worry about the nay say ers and talk of models. Let someone build it for the real world comparisons. I know you are busy but IF you have a little time to spare, I think several BUILDERS would like to give it a shot.

Thanks for ALL of your hard work and dedication to this thread...Thanks again.

 

[still4given]

I have a drawer full of Lateral and Vertical MOSFETs. I would need gerbers though. I don't want to invest that many devices on a home etch.

 

[jkuetemann]

I'd say a 3P or 5P would be the way to go, if you are going to go for the gusto and make the best possible do not cheap out or compromise. The EC output stage looks interesting and may well offer the overall best performance. My only issue is that lateral FETs have a different pin out and would make for a messy universal board. It is either a dedicated BJT and vertical FET board or a lateral FET design. Thoughts?

 

[still4given]

I have already built three pair of 5P BJT, one pair 2P BJT and a pair of 5P vertical Slewmaster OPS. I also built the little mini lateral OPS. What I'm hoping for here is something different and hopefully better though I'm not convinced that will happen. The OPS we have now are no slouches.

 

[MiiB]

I saw that DX-corp was working with an OPS, that has a diamond configuration around th drivers, would that yield any benefits..?

 

[vzaichenko]

I saw that DX-corp was working with an OPS, that has a diamond configuration around th drivers, would that yield any benefits..?

Well, this is a EF3 with the drivers, loaded each with individual CCS, ensuring the drivers operate in class "A" - approach, widely used by Electrocompaniet.
I tested it in one of my symmetric VFA designs, and it's definitely a good one - January 2014, you can see it here - >XHS Symmetric<.
I did not make direct comparisons with traditional EF3 or any MOSFET options, but it's stable, low distortion and good sounding.
Potentially lower crossover distortion, but this needs some more "research".

I slightly modify that prototype now (with regards to its IPS - came up with interesting approach to CMCL - need to test it), but will try to do some measurements to see if we can really benefit from those CCS'ed drivers in the nearest days.

Cheers,
Valery

 

[Esperado]

Ok, i tried to add driver to the DVSSA. It makes a DDSSA (Diamond driver simple symetric amplifier).
Indeed, it give huge improvement. at 60W, i was able to sim (with cordel models ;-)
1KHz hd: 0.000255%
10KHz hd: 0.000704%
Slew-rate (passive filters out) : 2320V/µs bandwidth: flat to 20MHz.
(Who says "slewmaster" ?

But square waves overshoot, and, oh, my god, what a rock'nroll of phases at the top of the bandwidth !

Hard work to try to civilisate the monster. No more overshoot:
1KHz: 0.000260%
10KHz: 0.000879%
Who need hundred transistors ?
Looks cool ? I will never build such an amp with all those evils at 100MHz !!!
I'm sure it will not sound so good than his little brother: one pole too much in the loop, too tricky to compensate..
http://www.esperado.fr/vssa-diamond/DDSSA.jpg (square waves are at 50KHz)

 

[jwilhelm]

On a Lateral output board with 5 pairs of devices would there be any advantage to running source resistors besides current monitoring for a protection circuit? Would they still balance the loads like in a BJT circuit? Would the Slewmaster Mini driver circuitry work or would it need some modification for biasing extra devices?

 

[jkuetemann]

My understanding that source resistors are not generally used because they have to be rather high in resistance to actually be useful in enforcing current sharing. Better to buy a larger sample of devices and match them.

Select drivers such that they can run at 25mA or so as a minimum (or at least 5mA per output pair), or higher if the dissipation is kept acceptable. A single EF driver like the Mini should be fine but you could use an EF2 if you really want added isolation from VAS loading.

Since laterals don't really need thermal compensation when idling in the 100mA to 150mA range you then only need to compensate for driver temperature.

 

[jwilhelm]

I think I'll still run 0R1 source resistors for current sensing. Any idea how close devices are as of late? Will I get 5 matches out of 10?

 

[jkuetemann]

That I can't say as I have only used them as single pairs. Mind you that you only match the N channel devices together and the P channel to themselves, there is no need to try and match the Ns and Ps to one another.

I would use the 0R1 as you suggest, if for no other reason than for current sensing. Bear in mind that since laterals have some self protecting characteristics that protection can be more basic and set to activate at somewhat higher levels of abuse than with either BJTs or VMOS.

I will have to see how many pairs I have left and see what they look like on the curve tracer, maybe tomorrow.

 

[jwilhelm]

Where did you find your laterals for the VSSA? I can special order them through Digikey. I haven't found them anywhere else.

 

[jkuetemann]

I get them from Tayda Electronics. Best price I've found and they seem to be genuine Renesas devices.

 

[jwilhelm]

I saw them but thought they looked a little sketchy. I'll check them out again.

 

[gaborbela]

I just finished the layout for the 3 pair lateral, it was posted at #4465.
These was designed by Cordell if I'm correct.
These is website.
Schematics
I read lost of good things about these PS.
My question someone with experience willing to double check the layout if is correct.
I do not post here, respect to OS work!
I did asked Jason if these will work with CFA-XH the answer was yes..
If you send an e-mail address I will send it to you.
I do not want to start a new thread for these.
These layout was drawn by hand so there is no silk screen but the same time all the parts are marked.
Thanks in advance

 

[jwilhelm]

I just finished the layout for the 3 pair lateral, it was posted at #4465.
These was designed by Cordell if I'm correct.
These is website.
Schematics
I read lost of good things about these PS.
My question someone with experience willing to double check the layout if is correct.
I do not post here, respect to OS work!
I did asked Jason if these will work with CFA-XH the answer was yes..
If you send an e-mail address I will send it to you.
I do not want to start a new thread for these.
These layout was drawn by hand so there is no silk screen but the same time all the parts are marked.
Thanks in advance

That looks pretty close to what I have drawn up already. I'm just reworking a Slewmonster 5P board so I can drop it in place of a BJT board. I want to try it with one of Vzaichenko's tube inputs.

 

[still4given]

The only amp I've built so far that had multiple LFET's was the ESP P101. It only has two pair of outputs and Rod Elliot said I must use matched devices. I'm not sure how closely they need to be matched but my understanding is that if they are not one device could end up hogging all the current.
The P101 uses 0R33 source resistors and 560R gate stoppers.

 

[jwilhelm]

I think in theory they should self balance at full power but I could see one conducting more at low power levels which would defeat the purpose of multiple devices. Hopefully they come in as closely matched as BJTs. Otherwise I'll have lots of outputs for VSSAs and Slewbabies.

 

[Esperado]

On a Lateral output board with 5 pairs of devices

Please, consider that Laterals like to run at 100-150mA of quiescent.(Each !)
Do you really want 500-750 ma ?
With multiple units, you will probably need more current margin, because of their disparities if not perfectly matched.
Parasitic gate capacitances will probably increase in such a arrangement, making them harder to drive at HF.
Personally I've never used more than two laterals in my amps.

 

[MiiB]

Goldmund uses several pairs of lateral Fets in their amps. They run them with quite low quiescent current. Think i once saw a thread here with the schematics.