I have been asked about Fuse ratings for the PSU circuit published in post #109.
Under normal operation, you have a total power consumption of 330W for 2 channels F5X plus protection.
So for those with 115Vac mains, 3A slow-blow is definitely too low. You might wish to try 6A slow-blow.
This should also cover the transient inrush current during power-up, which is limited by the Ametherm MS32 20010 NTC.
For people using 230Vac mains, a 3A slow-blow would have been OK.
But the inrush can be as high as 11A even for a very short period of time.
So you may reconsider replacing the MS32 20010 with a MS32 50006L, or MS22 75004.
Ametherm MS32 50006-L ---50 ohm / 6 Amp Inrush Current Limiter Data Sheet
Ametherm MS22 75004 ---75 ohm / 4 Amp Inrush Current Limiter Data Sheet
In that case, the max. inrush current is 4.5A and 3A respectively.
A 3A or 5A slow-blow would be the one I try first.
We'll put up a revised schematics for the power supply later.
Patrick
Under normal operation, you have a total power consumption of 330W for 2 channels F5X plus protection.
So for those with 115Vac mains, 3A slow-blow is definitely too low. You might wish to try 6A slow-blow.
This should also cover the transient inrush current during power-up, which is limited by the Ametherm MS32 20010 NTC.
For people using 230Vac mains, a 3A slow-blow would have been OK.
But the inrush can be as high as 11A even for a very short period of time.
So you may reconsider replacing the MS32 20010 with a MS32 50006L, or MS22 75004.
Ametherm MS32 50006-L ---50 ohm / 6 Amp Inrush Current Limiter Data Sheet
Ametherm MS22 75004 ---75 ohm / 4 Amp Inrush Current Limiter Data Sheet
In that case, the max. inrush current is 4.5A and 3A respectively.
A 3A or 5A slow-blow would be the one I try first.
We'll put up a revised schematics for the power supply later.
Patrick
Last edited:
Patrick,
Would you care mind explainaing the importance of the variable current on the lab supply. Variable voltage is self explanantory, but i am not quite sure of the purpose of the current regulation. Is it to try to match the current level thta the fets would see from the input Jfets under normal operating conditions? Also, should the fets be tested with source resistors in place. I will not be doing the oven trick, instead going the slow route of letting them heat up on a heatsink.
I said in that article that you should use a power supply with an adjustable current limit. By that I mean you get set the max. current, and not constant current. When you are testing a FET at say 4A, you probably want to set the current limit at say 4.5A, so that even if you have accidentally applied too much Vgs, or the FET has a high tempco, you are still safe. It is purely a precautionary measure.
It is not necessary to measure with the source resistor in place.
Hope this answers your question.
Patrick
It is not necessary to measure with the source resistor in place.
Hope this answers your question.
Patrick
Last edited:
GBs
This project is fantastic for a relative novice like me and I really appreciate the efforts put in by the team and everyone else who has helped out. That includes the organisers of the GBs, which are also a big help. However, I suspect I am not the only person who has struggled to keep up to speed with the multiple GBs. I therefore thought it might be useful to try and capture in one place brief details of the GBs and the current status. So here is an initial stab at this:
Toroidy trafos: shipped
Fuku resistors/fets/keratherm pads: shipped
Case & heatsinks: pending
CRC PCBs + MPC74 resistors: pending
2 amp boards + 8 cap boards + protection board: pending
Protection board parts kit + spares + small parts kit: pending
Matched output fets: trying to source?
IEC connector : pending
If I have got any of the above wrong or missed something then please amend it and re-post.
This project is fantastic for a relative novice like me and I really appreciate the efforts put in by the team and everyone else who has helped out. That includes the organisers of the GBs, which are also a big help. However, I suspect I am not the only person who has struggled to keep up to speed with the multiple GBs. I therefore thought it might be useful to try and capture in one place brief details of the GBs and the current status. So here is an initial stab at this:
Toroidy trafos: shipped
Fuku resistors/fets/keratherm pads: shipped
Case & heatsinks: pending
CRC PCBs + MPC74 resistors: pending
2 amp boards + 8 cap boards + protection board: pending
Protection board parts kit + spares + small parts kit: pending
Matched output fets: trying to source?
IEC connector : pending
If I have got any of the above wrong or missed something then please amend it and re-post.
I understand there is no place to ask this, but please try to keep this for technical discussion only if possible.
> Case & heatsinks: pending
> CRC PCBs + MPC74 resistors: pending
They were shipped together, in bulk. And they have arrived in Austria.
Please contact smyslow for details.
> 2 amp boards + 8 cap boards + protection board: pending
At Austrian Customs waiting to be handled, AFAIK.
> Protection board parts kit + spares + small parts kit: pending
You paid 2 days ago. Please give Mark 2 weeks.
> Matched output fets: trying to source?
You as a group need to get organised. Nikon1975 will help to cource.
But someone needs to take the lead.
> IEC connector : pending
Not 100% sure, but I think also with the case parts.
Patrick
> Case & heatsinks: pending
> CRC PCBs + MPC74 resistors: pending
They were shipped together, in bulk. And they have arrived in Austria.
Please contact smyslow for details.
> 2 amp boards + 8 cap boards + protection board: pending
At Austrian Customs waiting to be handled, AFAIK.
> Protection board parts kit + spares + small parts kit: pending
You paid 2 days ago. Please give Mark 2 weeks.
> Matched output fets: trying to source?
You as a group need to get organised. Nikon1975 will help to cource.
But someone needs to take the lead.
> IEC connector : pending
Not 100% sure, but I think also with the case parts.
Patrick
Zhou also raised the price for N-P match. I think his N-P match yield of his
latest match (whatever the method is) was quite low.
Since he bought a lot, I suspect it will need a very large pool to come up with
enough N-P matches. My random 20 pairs from a US supplier don't even have any N-P remotely close to within Vgs 0.25V. Out of the 0 pairs. Since all N's and P's are of same date code with respect to the type, this makes me think, if you are bad luck and purchase a bulk of P's that are grossly different from the N's, then a good portion of the lot may not be useful for N-P match at all!!
Patrick mentioned that the N-N and P-P matches are more important as they make up the 2 sides of the balanced channel, it will be interesting to know how bad it will be if N-P are totally unmatched, but the N-N and P-P are.
latest match (whatever the method is) was quite low.
Since he bought a lot, I suspect it will need a very large pool to come up with
enough N-P matches. My random 20 pairs from a US supplier don't even have any N-P remotely close to within Vgs 0.25V. Out of the 0 pairs. Since all N's and P's are of same date code with respect to the type, this makes me think, if you are bad luck and purchase a bulk of P's that are grossly different from the N's, then a good portion of the lot may not be useful for N-P match at all!!
Patrick mentioned that the N-N and P-P matches are more important as they make up the 2 sides of the balanced channel, it will be interesting to know how bad it will be if N-P are totally unmatched, but the N-N and P-P are.
The team has not done enough examples of the F5X to tell you how important.
Suffice to say that in the original, single ended F5 with totally unmatched Fairchilds (by nature), the even harmonics are still OK.
So I would not be totally worried if N-P match is not perfect.
The left-right balance cancellation will save the day.
The test team actually built one proto with 2SK3497 / 2SJ618 which are, by nature, not complementary.
The measurements are not worse than fully NNPP matched K1530/J201.
Have not yet had a chance to listen properly (that team member fully loaded with work).
Maybe you should just buy a pair of matched Ns and a pair of matched Ps from Zhou and test those with your first build.
Changing 4 power FETs later (if at all) is not a big operation.
Certainly much cheaper, considering his current asking price.
Now you see why I always build balanced (apart from constant current draw in Class A).
Patrick
Suffice to say that in the original, single ended F5 with totally unmatched Fairchilds (by nature), the even harmonics are still OK.
So I would not be totally worried if N-P match is not perfect.
The left-right balance cancellation will save the day.
The test team actually built one proto with 2SK3497 / 2SJ618 which are, by nature, not complementary.
The measurements are not worse than fully NNPP matched K1530/J201.
Have not yet had a chance to listen properly (that team member fully loaded with work).
Maybe you should just buy a pair of matched Ns and a pair of matched Ps from Zhou and test those with your first build.
Changing 4 power FETs later (if at all) is not a big operation.
Certainly much cheaper, considering his current asking price.
Now you see why I always build balanced (apart from constant current draw in Class A).
Patrick
PS :
Andrew is of course correct.
The way that Zhou matches is not truely F5X working conditions.
So even a perfectly matched N-pair with his one-point match at 500mA does not guarantee proper curve tracer match over operating range of F5.
That was why I proposed you GB your own random MOSFETs and then match under my instructions / using my equipment.
Even a 3-point match as I proposed will do.
Patrick
Andrew is of course correct.
The way that Zhou matches is not truely F5X working conditions.
So even a perfectly matched N-pair with his one-point match at 500mA does not guarantee proper curve tracer match over operating range of F5.
That was why I proposed you GB your own random MOSFETs and then match under my instructions / using my equipment.
Even a 3-point match as I proposed will do.
Patrick
I think the problems are the following:
1) In order to get fairly good matches we would probably need something like 500 + 500 original virgin devices at a cost which would most likely not be inferior to 5$/device (i.e. 5.000$ total).
2) If this cost was to be divided by subscribers (n = 25) it would be 200$ per/F5X amp.
3) As only a fraction of the devises would be used, who should benefit financially from the left-overs? I'm sure the devices will maintain their value in time, but this is difficult to quantitate now.
4) Many (most?) of us already have "poorly" matched devices so why spend another 200$ at the gain of what?
5) Who would possibly have the time and measures to match the devices well in a fairly short time period?!
All other parts of the F5X build should be with the end-users in a month or so.
I don't have a solution. I would be ready to invest financially in the endeavor, but I absolutely do not have the time to do the matching or the F5X might be my last amp while having a job......
Anybody willing to spend a few weeks on matching?
Cheers,
Nic
1) In order to get fairly good matches we would probably need something like 500 + 500 original virgin devices at a cost which would most likely not be inferior to 5$/device (i.e. 5.000$ total).
2) If this cost was to be divided by subscribers (n = 25) it would be 200$ per/F5X amp.
3) As only a fraction of the devises would be used, who should benefit financially from the left-overs? I'm sure the devices will maintain their value in time, but this is difficult to quantitate now.
4) Many (most?) of us already have "poorly" matched devices so why spend another 200$ at the gain of what?
5) Who would possibly have the time and measures to match the devices well in a fairly short time period?!
All other parts of the F5X build should be with the end-users in a month or so.
I don't have a solution. I would be ready to invest financially in the endeavor, but I absolutely do not have the time to do the matching or the F5X might be my last amp while having a job......
Anybody willing to spend a few weeks on matching?
Cheers,
Nic
I am sure most of you would want spares.
Each of you would need 8 FETs. So 240 for 30 people, plus one set spares.
We are already 500. Matching reject rate 50%, and you have your 1000 devices for 5000 Euros.
And yes, same price as a set from Zhou, but properly matched.
(And also yes, he has a large pool to match N-P, be it at 500mA.)
Again yes, if most have already matched devices then little point.
But I was watching the list on the GB thread for matched devices, and the interest was not small.
http://www.diyaudio.com/forums/grou...e-user-distribution-check-16.html#post2895477
Matching is the problem I cannot solve for you.
I am prepared to spend a few hours coaching, and let you borrow my equipment.
The rest you need to organise.
Ideally someone in South of Germany should do the matching.
But your call. I was just trying to help.
I am happy either way, as I don't gain anything other than that the F5X will be built as best as we possibly can.
Patrick
Each of you would need 8 FETs. So 240 for 30 people, plus one set spares.
We are already 500. Matching reject rate 50%, and you have your 1000 devices for 5000 Euros.
And yes, same price as a set from Zhou, but properly matched.
(And also yes, he has a large pool to match N-P, be it at 500mA.)
Again yes, if most have already matched devices then little point.
But I was watching the list on the GB thread for matched devices, and the interest was not small.
http://www.diyaudio.com/forums/grou...e-user-distribution-check-16.html#post2895477
Matching is the problem I cannot solve for you.
I am prepared to spend a few hours coaching, and let you borrow my equipment.
The rest you need to organise.
Ideally someone in South of Germany should do the matching.
But your call. I was just trying to help.
I am happy either way, as I don't gain anything other than that the F5X will be built as best as we possibly can.
Patrick
Last edited:
I have been asked by PM about sourcing MOSFETs.
You probably all know that Toshiba is stopping all audio devices, MOSFETs and JFETs alike.
The only open source that is known to have sufficient quantities for NNPP match is, as mentioned before, ZhouFang.
It has also been discussed before that his matching is not exactly at F5X operating conditions.
So if you do not wish to purchase from Zhou, the only solution right now is to GB and match yourselves.
We ourselves are in the process of acquiring sufficient devices to support future batches of F5X projects.
But it won't be 6 months or so before we are ready, and then only in the context of supporting the project,
and not as separate devices sets for general sales.
Patrick
You probably all know that Toshiba is stopping all audio devices, MOSFETs and JFETs alike.
The only open source that is known to have sufficient quantities for NNPP match is, as mentioned before, ZhouFang.
It has also been discussed before that his matching is not exactly at F5X operating conditions.
So if you do not wish to purchase from Zhou, the only solution right now is to GB and match yourselves.
We ourselves are in the process of acquiring sufficient devices to support future batches of F5X projects.
But it won't be 6 months or so before we are ready, and then only in the context of supporting the project,
and not as separate devices sets for general sales.
Patrick
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.