o.k. today...
I looked for the distortion spectra of my BA-3/crippled (only one pair of IRFs) F4 combination,
and the k2 was very dominant, about 20db higher than k3, THD was around 0.034%.
That was a surprise, the J-Fets of the BA-3 input were matched in Idss and the F4 output stage itself was k3 dominant. So I expected from the input stage also k3 dominance.
I measured both Idss of J74 and K170 again, values were 9,64mA/9,62mA. not bad....nevertheless....
Conclusion is, same Idss seems not to be a guaranty for a k2 minimum and k3 dominance....
I wanted to respond with this yesterday http://www.diyaudio.com/forums/pass-labs/121228-f5-power-amplifier-174.html#post2277700
but, I couldn' find it then...
Idss matching does not allow for Matched S. This article shows a methode to get closer Matching
Thanks!
I got by experiment with P3 the same resistor imbalance of around 5R Euvl mentions....To get lowest distortion.
If I remember right at the moment......
That means without this correction the higher transconductance of the J74 with the same Idss as the K170 will produce k2 with positive Phase, mostly wanted...
Am I right?
I got by experiment with P3 the same resistor imbalance of around 5R Euvl mentions....To get lowest distortion.
If I remember right at the moment......
That means without this correction the higher transconductance of the J74 with the same Idss as the K170 will produce k2 with positive Phase, mostly wanted...
Am I right?
Depends on the imbalance in the output. IF you minimize the imbalance in the FE, then it just shifts to the output. In theory, you could do somethning similar there with Rs values, but if you like, you like. I will be going quazy in matching the output fets on the next version. I will match for Vgs, Yfs, and Ciss. I have a pool of about 50 matched at Vgs to pull from for each polarity device.
jFETs that are selected for same Idss are NOT MATCHED, they are simply Idss selected.
Matching is a very much more involved parameter measuring and matching procedure.
Friends please tell me...
why are Mosfets matched for Vgs at a certain Id
and why are J-Fets matched with Idss and not like MosFets.....?
As Andrew points out, there are several parameters to select for before we should actually be calling these devices complimentary match's or same sex match's.
When parraleling MOSFETs ussually you need to drive them with equal Vgs, as they are in parallel, and this would allow for equal current sharing if matched. In the case of comp devices the situation is similar but if N-P can't be matched your drive circuit will have an offset in order for the output to have 0V offset. If the N-P are expected to be complimentary devices, matching 1 parameter should be a close match for the others. In both cases the Id we shoot for is our typical operating point.
I don't really have a good answer for the JFET matching philosophy other than it is easyest to match JFETs at 0Vgs > Idss. Sometimes we use them at that 0Vgs like maybe the B1. Again, as Andrew says, we are only selecting parameters that are easier to measure expecting the rest to follow suit.
Pass-Out then ?
(matching pants not only implies identical color and fabric, but also width, length, stitches, just not a faked brand label)
It lives...
I don't know if I'm proud or embarrassed but I got to give a listen. It took a while to make the mosfets behave but it seems to be working well. Two pairs of buz 901/906 driven by ba3 front end. The bias is set at 1.6 amps per output fet.
crclc power supply with 400 va 18 volt transformer.
I plan to build three more of the output stages and try them paralleled and balanced.
I can't really comment on the sound listening to one channel, but first impressions are pretty good.
Anyway it's fun to make stuff and even more fun when it works...
E
I don't know if I'm proud or embarrassed but I got to give a listen. It took a while to make the mosfets behave but it seems to be working well. Two pairs of buz 901/906 driven by ba3 front end. The bias is set at 1.6 amps per output fet.
crclc power supply with 400 va 18 volt transformer.
I plan to build three more of the output stages and try them paralleled and balanced.
I can't really comment on the sound listening to one channel, but first impressions are pretty good.
Anyway it's fun to make stuff and even more fun when it works...
E
Attachments
Finally got two channels going...Holy Sh!t this thing sounds good. There is so much different from other builds it's hard to say where the improvement is coming from. This is the second time I've used buz 901/906 outputs and I think that they may be part of what I like. The amp is very snappy. I will be building a couple more sets of output blocks to try balanced as well as well as just paralleling more outputs.
Thanks again too all the people who asked questions and answered them before I even had them...
Evan
Thanks again too all the people who asked questions and answered them before I even had them...
Evan
Attachments
