Krell KSA 100mkII Clone

[TomWaits]

Originally posted by GeWa
GUESS WHAT WAS IN THE MAILBOX TODAY!!!

Good for you man. Mine showed up in the mail wrapped like English Fish & Chips. The wrapping paper, that is.

They are indeed very sweet looking.

Cheers,

Shawn.

My boards are just fine, BTW. Cheers to the married couple!

[analog_guy]

I have read several suggestions regarding the substitution of IR MOSFETS for the VN/VP devices found in such amps as the Krell.
Before making any such substitution, I strongly advise that the circuit first be simulated. In particular, the transfer characteristics of most switching MOSFETs (and that is what IR devices are optimized for) differs considerably from VN/VP devices, particularly for low Vgs. When simulating, consider using a two-tone input signal and examine the resulting spectrum at the output.

The VP/VN devices are low threshold MOSFETS, with a Vt in the 1.0 - 2.0 volt range, as compared to the 5-10V Vt specified for switchmode MOSFETs. My experience with switching MOSFETs is that they require a high Vgs (read high quiescent current) to guarantee linear operation, and that requirement may conflict with other circuit requirements, especially if used in the input or VAS stages of an amplifier.

By the way, Supertex (I believe they were ther original mfgr) is still in business and sells a good selection of low Vgs MOSFETs. Additionally, some are second sourced by Zetex and are available from DigiKey.

Hope this helps,

JCM

[PWatts]

True indeed. That was also partly the reason I made provision for three different packages for the mosfets to provide as much flexibility as possible.

I have SPICEd the circuit extensively before laying out the board, and indeed the lower Vt ones performed better since the bias current isn't too high in this application. That is also why the preferred choice for the amp is specified as ZVN/ZVP2110 since it's not a dedicated switching type and has a Vt of 0.8-2.4V compared to the 2-4V of the IRF ones.

However, Mark tested his amp with both types and could not notice any difference, and I plan to do a full set of Audio Precision experiments when I'm done building. The lavish Chinese clone even used IRF610/9610 which has even higher Vt (not a problem for Pass applications since they're heavily biased).

Flodstroem:

I haven't received the boards yet so have sadly not begun building, although besides you it seems like nobody else has either. Funny how all those boards got sold so quickly and people are already discussing KMA160's Hopefully that will change once everyone has their diodes. I have done some prepping work however by matching all the transistors & zeners, milling heatsinks and designing and building a prototype for a front-end regulator. Fortunately the first amp (I'm building 3) is merely an upgrade from a commercial Mk1 to Mk2 so it's only swopping out the boards which should go quick(ish).

[Flodstroem]

Originally posted by analog_guy
I have read several suggestions regarding the substitution of IR MOSFETS for the VN/VP devices found in such amps as the Krell.
Before making any such substitution, I strongly advise that the circuit first be simulated. In particular, the transfer characteristics of most switching MOSFETs (and that is what IR devices are optimized for) differs considerably from VN/VP devices, particularly for low Vgs. When simulating, consider using a two-tone input signal and examine the resulting spectrum at the output.

The VP/VN devices are low threshold MOSFETS, with a Vt in the 1.0 - 2.0 volt range, as compared to the 5-10V Vt specified for switchmode MOSFETs. My experience with switching MOSFETs is that they require a high Vgs (read high quiescent current) to guarantee linear operation, and that requirement may conflict with other circuit requirements, especially if used in the input or VAS stages of an amplifier.

By the way, Supertex (I believe they were ther original mfgr) is still in business and sells a good selection of low Vgs MOSFETs. Additionally, some are second sourced by Zetex and are available from DigiKey.

Hope this helps,

JCM

Originally posted by PWatts

Flodstroem:
I haven't received the boards yet so have sadly not begun building, although besides you it seems like nobody else has either. Funny how all those boards got sold so quickly and people are already discussing KMA160's Hopefully that will change once everyone has their diodes. I have done some prepping work however by matching all the transistors & zeners, milling heatsinks and designing and building a prototype for a front-end regulator. Fortunately the first amp (I'm building 3) is merely an upgrade from a commercial Mk1 to Mk2 so it's only swopping out the boards which should go quick(ish).

analog_guy, I totally agree with your comments here. This was the reason for the heavy check for substitutes for the old obsolete Supertex VN/VP0210N5 (To-220 devices). I have found several types of Vertical DMOS that could be useful for this application, but best fit was the Zetex ZVN/ZVP 2110G-types (SOT-223) what I can see after studying and reading old and new data sheets carefully. Only (and slightly) difference was in the Source resistance and of course in the power dissipation-factor). When matching those Zetex I could se that the Vt was in the 1.0 - 2.0 volt range, as you said (mine measured in the range of 1.1-1.36V)
Regarding simulation: how do you simulate with those old obsolete Supertex V-DMOS devices. Do you need some Spice model/data for to do it or is it enough with the Krell KSA100-schematics?
Could you do a simulate with those Zetex to find out if they really are an alternativ?

BTW, I have received 70pair as samples of those V-DMOS from Zetex, not bad at all. I newer received any 1N5309 samples and that was not so good, I had to order those from Mouser. Bad luck Zetex dont make such a diodes.

PWatts, Yes I have been matching devices for the amp too, eg. all semiconductors except for the powers. Boards are nearly finished (6 boards) but as I said earlier, I lack the heat sink for the drivers (Q14-Q19, where 18 and 19 are mine extra pair). Also not soldered is the 12pcs CP-optional caps.
Im wandering if there is somebody sitting with a huge lot of Panasonic FC-caps for this application? If there is, maybe you would like to exchange some for to get Zetex ZVP/ZVN V-DMOS Fets, or just sell a batch to me? Send PM please.

Hope you will receive the boards soon, due to I am very curious about if you could do some measurements and distortion analyze?

Regards

[loek]

Hi Flodstroem, can you be more specific which ones?
The 10uF is not easy in 2mm pitch, swap for 22uF/25V and those are available at RS. How many of each type?
I can order at RS and sent you if this may help.
regards, Loek

[PWatts]

Neither the value nor quality or ESR of the 10uF's are very critical. In fact I think a normal-ESR type would be a safer choice. Low ESR for the feedback caps are wasteful, rather go for low ESL (Elna/BG). For the PSU caps low-ESR would be a good choice though.

[Mark A. Gulbrandsen]

Pierre,
You guys should be seeing your stuff pretty darn soon... It left here the same day as GeWa's did. Let me kow when it arrives....


Flod... Any news on getting toroids made up? I am at a point that somethings gotta happen with it or I'll have to go another route.
Thanks!

Mark

[Flodstroem]

Originally posted by loek
Hi Flodstroem, can you be more specific which ones?

Originally posted by Mark A. Gulbrandsen
Flod... Any news on getting toroids made up? I am at a point that somethings gotta happen with it or I'll have to go another route.
Thanks!
Mark

loek, what I meant was the CP # 3 to #10 (470-1000uF/63 or 100V). Also I could use 220-330uF/100V Panasonic, Nichicon or Elna for the CP3, CP4. But for the pre-driver and input stage I would rather use the 560 or 680uF/63V.

If using the 220uF/100V combined with the 680uF/63V (Panasonic FC series or Nichicon UPW-series) you could get them in the hight of 20 mm and with a diameter of 18 mm (yes they fit on the board) this will prevent the board components from rising to much over the board and give you an easier fit in the chassis if it should be an issue.

Mark, yes I can understand your question, I have been waiting for it. Due to the low response I have only 6 pcs to order (your 2 pcs included). The price is pretty high as you know so thats the reason for the delay. If you really want to go for it and place a 2 pc order I could be temted to order my 4 pcs too.

If searching for other ideas, what alternatives do we have? What I can understand is that the only alternativ for the original x-former would be a custom design unit with the same specs (as best can) as for the original. Looking at other types of x-formers like the Plitron it could have been an alternativ if it wasnt for the expensive freight costs from Canada.

A custom design unit (800-900VA) would be a cheaper alternative to the original but not much. But maybe this is a way to go due to it could be more interesting to those that didnt afford to buy the originals. If I could get some inputs here from you guys who is going to build this amp, it could make it easier to do a final decision?

Price picture for an original Krell x-former would be approx £ 145 (6 or more pcs) and for a custom made approx £ 110. Both models will be potted in a mild steel can (black)
Remember, you will need two pcs for a stereo amp. Total VA to feed a stereo amp would then be approx. 1.8kW.
If designing a custom unit we could also use the same unit for the KAV-250a model when this time comes.

Further information could be found in post # 842-krell original x-former data, post # 852 - suggestion/discussion for a custom design unit.

Regards

[acenovelty]

Check these folks for the toroids:

http://www.toroid-transformer.com/

[Mark A. Gulbrandsen]

600 VA ANTEC is way too small for this amp. It might suffuce for an 8 ohm or higher load... The original can drive very low impedances quite easily.

Mark