So, I came across a pair of P-channel MOSFETs, the IRF4905, and I'm wondering if I could match them with some IRF510's I have for use in an ACA Mini clone I'm gathering parts for. While the Vgs range for both devices is similar (-2 to -4 for the IRF4905, 2 to 4 for the IRF510; I'll measure and try to match them), some of the other specs, particularly the transconductance, are scary different:
IRF4905
Rds(on) = 0.02R
Gfs = 21S
IRF510
Rds(on) = 0.54R
Gfs = 1.3S
The IRF610 seems an even worse match:
Rds(on) = 1.5R
Gfs = 0.8S
So could I modify R8 and R9 to degenerate these two better, say put a 1R3 on the IRF4905 and the spec 0R75 the IRF510? Or is this whole thing just ignorant folly given the Gfs gulf?
IRF4905
Rds(on) = 0.02R
Gfs = 21S
IRF510
Rds(on) = 0.54R
Gfs = 1.3S
The IRF610 seems an even worse match:
Rds(on) = 1.5R
Gfs = 0.8S
So could I modify R8 and R9 to degenerate these two better, say put a 1R3 on the IRF4905 and the spec 0R75 the IRF510? Or is this whole thing just ignorant folly given the Gfs gulf?
As I don't have access to the NP preferred Harris Mosfets, I was wondering if one could use the Toshiba 2SJ313 / 2SK2013 instead. I have some genuine NOS ones of these devices. If so, any changes recommended to resistor values?
I'm wrong if I buy IRF9520 and IRF520 from mouser without match them?
LSK170 and LSJ74 the only way is buy matched from diyaudiostore?
Any other cheaper solution?
Thank you
LSK170 and LSJ74 the only way is buy matched from diyaudiostore?
Any other cheaper solution?
Thank you
You can build these without part selection, but the performance will likely
be different. You can get 2SK170/2SJ74 from punkydawgs on ebay, genuine
matched, but you won't save any money. LSK170 and LSJ74 are similar
quality but I don't see better prices for them elsewhere at low quantities.
The IRF9520 and IRF520 are matched NOS Harris parts, and preferred, but
you can make it out of other Mosfets if you are prepared to adjust some of
the resistor values and such.
Lots of "buts". All told, I think the diyAudio offering will be reasonably priced.
be different. You can get 2SK170/2SJ74 from punkydawgs on ebay, genuine
matched, but you won't save any money. LSK170 and LSJ74 are similar
quality but I don't see better prices for them elsewhere at low quantities.
The IRF9520 and IRF520 are matched NOS Harris parts, and preferred, but
you can make it out of other Mosfets if you are prepared to adjust some of
the resistor values and such.
Lots of "buts". All told, I think the diyAudio offering will be reasonably priced.
Regarding matching, are LSJ74 and LSK170 (and, along the same lines, IRF9520 and IRF520) matched with their complements as well as each other? Or are they only matched with like devices?
The Harris parts came in fairly narrow band, about 4.2 to 4.5 Vgs , so it was not difficult to get .1V matches P to N.
This is not critical, but good matching helps in delivering the same performance as the prototype.
This is not critical, but good matching helps in delivering the same performance as the prototype.
I'm wondering if the DIYA store will have an option to just buy a set of the required matched active devices - I'm getting a PCB via one of the GBs that KevinHeem is running.
One credible strategy is to build what you have today, with what you can source today, and find out how happy that makes you. Later, if and when the Store has full kits including matched pairs of Harris MOSFETs, available for sale, you can decide whether to build that version too. Such a plan reduces anxiety, and lets you start listening to music on an amplifier you built yourself, right away. Minimum waiting time.
Would SMPS noise be a problem if the power module was inside a chassis with the ACA Mini? I was considering something like the Mean Well IRM-90-24 or IRM-90-24ST modules.
The question is motivated more by a idea of using 4 such module for FirstWatt class amps like the F4-F8 that use 300VA power transformers with 18V secondaries. Using SMPS modules instead of linear power supplies would provide some interesting packaging options. Here is one:
The question is motivated more by a idea of using 4 such module for FirstWatt class amps like the F4-F8 that use 300VA power transformers with 18V secondaries. Using SMPS modules instead of linear power supplies would provide some interesting packaging options. Here is one:
Attachments
Thanks Mark and yes that would be an option were it not for the half dozen or so projects already in the queue, so, by the time I get around to this one I'm sure the kit will be available in the store.
Here is the spectrum analysis I measured at 1W.
Inputs transistors installed are LS 170/74 matched at 8.5mA Idss.
Output mosfets are Harris (maybe) 9520/520 which measured consistently at a VGS of 3.20V for all P channel and 3.50V the N channel (I had 10 of each). So they are only matched with .3V
Compared to Nelson's measurements they look pretty close.
I have some Vishay parts coming in tomorrow. Should be interesting to play with those.
Inputs transistors installed are LS 170/74 matched at 8.5mA Idss.
Output mosfets are Harris (maybe) 9520/520 which measured consistently at a VGS of 3.20V for all P channel and 3.50V the N channel (I had 10 of each). So they are only matched with .3V
Compared to Nelson's measurements they look pretty close.
I have some Vishay parts coming in tomorrow. Should be interesting to play with those.
Attachments
Last edited:
Of interest Kevin is the fact that your are measuring a Vgs that is 1V lower than what Nelson has measured for the Harris types. Maybe because you are measuring at a different current to what his test jig allows. At least as you say the results are consistent, as Nelson also mentioned being in a narrow band for each type.
Yes, I'm curios about that as well.
I tested the mosfets at 5mA using a 22ohm resistor, 15V power supply, and connections as outlined in Nelson's article on matching devices. I would just tweek the power supply a touch to get exactly 5mA.
Maybe Nelson might like to comment. Another possibility is the Harris batches he has are different to what you have - maybe years apart.
Years aside, I measure vgs at about 100 ma. The Harris parts come in around 4.2 to 4.5 v. However they were chosen for relatively low distortion.