Zhou's method gives you a single point of Vgs at a specific current. He is most probably not measuring at 2A as we want.
As the xen audio site, matching section shows you, a single point at any current does not guarantee match over entire range.
xen-audio
Also depending on how long the current is on, the FET heats up to a more or less extent, causing Vgs to change.
The single point match at bias is a prerequisite for curve tracer match. So unless you are extremely lucky (then you should go and buy lottery), two mismatched FETs at bias would have very little change of curve trace match later under operating conditions (Vds, bias, and temperature).
Patrick
As the xen audio site, matching section shows you, a single point at any current does not guarantee match over entire range.
xen-audio
Also depending on how long the current is on, the FET heats up to a more or less extent, causing Vgs to change.
The single point match at bias is a prerequisite for curve tracer match. So unless you are extremely lucky (then you should go and buy lottery), two mismatched FETs at bias would have very little change of curve trace match later under operating conditions (Vds, bias, and temperature).
Patrick
Not yet, John. NicMac has started a list in post #748, the last addition to it is in post #793.
Nikon1975 has indicated that he may source the parts in Japan.
Nikon1975 has indicated that he may source the parts in Japan.
^ thanks Gerhard
NicMac 8 pairs (2 stereo F5X)
CanAm Man 4 pairs (1 stereo FX5)
ramallo 4 pairs (1 stereo FX5)
oneaclock 4 pairs (1 stereo FX5)
andrewt 4 pairs (1 stereo FX5)
Studley 4 pairs (1 stereo FX5)
ashaw 8 stereo sets?
triode4 4 pairs (1 stereo F5X)
heinz1 4 pairs (1 stereo F5X)
Mravinsky 4 pairs (1 stereo FX5)
JBdV 8 pairs for 2 F5X
smyslow 4 pairs (1 stereo F5X)
APK 4 pairs (1 stereo F5X) May be 8 pairs depending on price
Horio 4 pairs (1 stereo F5X)
Westley 4 pairs (1 stereo F5X)
alazira 4 pairs (1 stereo F5X)
pmchoong 4 pairs (1 stereo F5X)
CeeVee 4 pairs (1 stereo F5X)
johnwmclean 4 pairs (1 stereo F5X)
How many have already bought from Zhou?
NicMac 6 pairs
andrewt
Studley
JBdV 9 pairs
alazira
johnwmclean 4 pairs
NicMac 8 pairs (2 stereo F5X)
CanAm Man 4 pairs (1 stereo FX5)
ramallo 4 pairs (1 stereo FX5)
oneaclock 4 pairs (1 stereo FX5)
andrewt 4 pairs (1 stereo FX5)
Studley 4 pairs (1 stereo FX5)
ashaw 8 stereo sets?
triode4 4 pairs (1 stereo F5X)
heinz1 4 pairs (1 stereo F5X)
Mravinsky 4 pairs (1 stereo FX5)
JBdV 8 pairs for 2 F5X
smyslow 4 pairs (1 stereo F5X)
APK 4 pairs (1 stereo F5X) May be 8 pairs depending on price
Horio 4 pairs (1 stereo F5X)
Westley 4 pairs (1 stereo F5X)
alazira 4 pairs (1 stereo F5X)
pmchoong 4 pairs (1 stereo F5X)
CeeVee 4 pairs (1 stereo F5X)
johnwmclean 4 pairs (1 stereo F5X)
How many have already bought from Zhou?
NicMac 6 pairs
andrewt
Studley
JBdV 9 pairs
alazira
johnwmclean 4 pairs
I can also tell you that we are trying to source them in large enough quantities to make sure we get good curve tracer match.
But this will take time, both in procurement and in matching.
If you cannot wait for us, then you can GB and get them sent to heinz1.
He is then allowed to borrow equipment from me for matching.
Among all of you, I think he has the least distance to travel.
Whether you will be faster this way remains to be seen.
Most likely less costly, but also a smaller pool to match from.
Patrick
But this will take time, both in procurement and in matching.
If you cannot wait for us, then you can GB and get them sent to heinz1.
He is then allowed to borrow equipment from me for matching.
Among all of you, I think he has the least distance to travel.
Whether you will be faster this way remains to be seen.
Most likely less costly, but also a smaller pool to match from.
Patrick
Last edited:
> The front panel is a beauty. But it is smaller than I expect.
You have received the entire 350mm that we promised from day 1.
Right ?
Patrick
Yup, but on picture 350mm looks much bigger.How many capacitors do we need for the PSU? According to the attached pdf, which I downloaded somewhere sometime ago, we will need 8 x 22,000 uF. Surely these can't fit in the case.
> How many capacitors do we need for the PSU?
Refer to the latest Schematics and Bill of Material published by fitzfish in the F5X thread.
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-13.html#post2836103
> we will need 8 x 22,000 uF. Surely these can't fit in the case.
There are enough photos on the Xen Audio website and at the F5X thread to demonstrate how they will all fit in.
e.g.
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-11.html#post2738581
Patrick
Refer to the latest Schematics and Bill of Material published by fitzfish in the F5X thread.
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-13.html#post2836103
> we will need 8 x 22,000 uF. Surely these can't fit in the case.
There are enough photos on the Xen Audio website and at the F5X thread to demonstrate how they will all fit in.
e.g.
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-11.html#post2738581
Patrick
Last edited:
I also bought 4 sets of ppnn quads.
I paid the premium price and I had confirmation that the quads were to suit the F5x.
My batch of Pch were marked up @ -2.0057Vgs to -2.0076Vgs
And of Nch @ 1.8852Vgs to 1.8912Vgs.
The average diff Nch to Pch is ~ 120mVgs
Euvls average diff Nch to Pch is ~ 5mVgs
Was I just unlucky that my sets were 24times worse?
I did not get what I ordered.
I did not get what I paid for.
I feel I was robbed.
> How many capacitors do we need for the PSU?
Refer to the latest Schematics and Bill of Material published by fitzfish in the F5X thread.
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-13.html#post2836103
> we will need 8 x 22,000 uF. Surely these can't fit in the case.
There are enough photos on the Xen Audio website and at the F5X thread to demonstrate how they will all fit in.
e.g.
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-11.html#post2738581
Patrick
I see. How big are these caps? Is there a GB for these?
I have 22,000uF 63V Nichicon Caps and they are huge. Wonder if I can use them.
What is the white piece of 'plastic' that holds the caps and are they included in the case?
> Was I just unlucky that my sets were 24times worse?
You forgot to point out that the N-N match we got were to 0.1V only.
Or 99mV if you prefer.
And that is from a pool 5 times larger than what you bought.
We bought ours very early on when I think he still had sets with good matches.
But only Zhou can answer that.
Or maybe others who bought from Zhou as well can measure and publish to give further indication.
I cannot say his FETs are outside matching spec, whatever he promised.
Because he did not specify matching at 2A.
I merely want to point out that at 2A (nominal bias of F5X), the match is not as well as the numbers indicated on the devices.
Also to give you an indication, we bought 50 pieces of each unmatched from a Japanese DIY supplier a few years back.
From those 100 pcs (50x P & 50x N), we did not find any with PN match to 0.1V.
Such is the difficulty of obtaining P-N match.
So even if you GB together, don't count on it that many of you will get perfect P-N match, whatever perfect means.
Maybe with luck to 0.1V.
Unless you buy thousands and cherry pick.
Patrick
You forgot to point out that the N-N match we got were to 0.1V only.
Or 99mV if you prefer.
And that is from a pool 5 times larger than what you bought.
We bought ours very early on when I think he still had sets with good matches.
But only Zhou can answer that.
Or maybe others who bought from Zhou as well can measure and publish to give further indication.
I cannot say his FETs are outside matching spec, whatever he promised.
Because he did not specify matching at 2A.
I merely want to point out that at 2A (nominal bias of F5X), the match is not as well as the numbers indicated on the devices.
Also to give you an indication, we bought 50 pieces of each unmatched from a Japanese DIY supplier a few years back.
From those 100 pcs (50x P & 50x N), we did not find any with PN match to 0.1V.
Such is the difficulty of obtaining P-N match.
So even if you GB together, don't count on it that many of you will get perfect P-N match, whatever perfect means.
Maybe with luck to 0.1V.
Unless you buy thousands and cherry pick.
Patrick
Last edited:
> I have 22,000uF 63V Nichicon Caps and they are huge.
> Wonder if I can use them.
No. All details including maximum size of caps have been thoroughly discussed and published.
Please go through the information in the thread. I can't do this for you.
Else just follow the part number in the Bill of Material published.
> What is the white piece of 'plastic' that holds the caps and are they included in the case?
Yes. Part of option 2.
Patrick
> Wonder if I can use them.
No. All details including maximum size of caps have been thoroughly discussed and published.
Please go through the information in the thread. I can't do this for you.
Else just follow the part number in the Bill of Material published.
> What is the white piece of 'plastic' that holds the caps and are they included in the case?
Yes. Part of option 2.
Patrick
Some more info on the importance of P-N match for the F5X can be found here :
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-14.html#post2903419
Patrick
http://www.diyaudio.com/forums/pass-labs/183362-f5x-euvl-approach-14.html#post2903419
Patrick
I just finished my measurement to Zhou's N-P match sets and my 10+10 random buy.
First, I can pretty much confirm that Zhou didn't match at 2A. It was more like at the 500ma range.
From Zhou, the 8 pair of devices have:
2SK1530 Vgs: Zhou's measurement, 1.9770 to 1.9705, diff 6.5ma
From my single point 2A matching, I got 2.453 to 2.388, diff 65ma.
2SJ201 Vgs: Zhou's measurement, 1.9772 to 1.9720, diff 5.2ma.
From my single point 2A matching, I got 2.419 to 2.411, diff 8ma.
Max N-P diff = 42ma.
From the 10+10 random buy:
2SK1530 Vgs: 2.465 to 2.382, diff 83ma.
2SJ201 Vgs: 2.357 to 2.216, diff 140 ma.
Max N-P diff: 249ma.
Out of the 10 pairs of random pick's, there are only 2 pairs of N-P that are matched (single point bias) within 0.1V which is probably not bad; however, the untold story is that the P-channels have a very wide range, which is very much what I observed when I matched other Toshiba devices. The N's are usually in a tighter range, but the P can go pretty wild. My gut feeling is that, it will need a good size pool to yield good P-matches.
I have to say that I might be at luck to get a reasonably close Octet's from a single bias matching point of view.
Fred,
Firstly all your readings are Vgs and hence they should be mV and not mA.
Secondly I think you have very good sets from Zhou.
For some reasons, his N-match is much worse.
Probably due to spread over a few days, or whatever that is not held constant.
And you are very lucky with you 10+10 random.
As I mentioned in post #825, if you all decide to GB 500 each and then group match, you will get help from me.
OK ?
Patrick
Firstly all your readings are Vgs and hence they should be mV and not mA.
Secondly I think you have very good sets from Zhou.
For some reasons, his N-match is much worse.
Probably due to spread over a few days, or whatever that is not held constant.
And you are very lucky with you 10+10 random.
As I mentioned in post #825, if you all decide to GB 500 each and then group match, you will get help from me.
OK ?
Patrick
As reported earlier, we had a sizing problem with the regulator heat sink.
As a result, we have shipped the original heatsink with a CRC PCB, and provided MPC74 resistors, at our costs.
This enables you to build the F5X using a CRC supply.
We are pleased to inform you that we have successfully tested a ultimo solution.
Thanks to Mark for all the testing efforts.
This solution uses 2x Fischer SK466-38 heatsink, bolted to a custom machined U-channel.
On the channel are mounted 8x TO220 rectifying diodes, as well as the original MOSFET regulator.
At regulator full load (output +/-16V 4A), the temperature rise from ambient was 32°C.
We are therefore happy enough to release this solution as it.
See attached picture.
For comparison the original heatsink is on the left, and the caps on top & right.
Patrick
As a result, we have shipped the original heatsink with a CRC PCB, and provided MPC74 resistors, at our costs.
This enables you to build the F5X using a CRC supply.
We are pleased to inform you that we have successfully tested a ultimo solution.
Thanks to Mark for all the testing efforts.
This solution uses 2x Fischer SK466-38 heatsink, bolted to a custom machined U-channel.
On the channel are mounted 8x TO220 rectifying diodes, as well as the original MOSFET regulator.
At regulator full load (output +/-16V 4A), the temperature rise from ambient was 32°C.
We are therefore happy enough to release this solution as it.
See attached picture.
For comparison the original heatsink is on the left, and the caps on top & right.
Patrick
Attachments
Last edited:
To implement this solution, you will need to GB, for each case 4x, SK466-38 heatsink extrusion from Fischer.
Probably best done by someone living in Germany, Holland, or Austria, preferably with a company.
There are two through holes required to bolt them onto the U-Channel.
If Fischer charges too much for the drilling, there is no reason you cannot drill yourselves.
Then you would need a conversion kit from us, consisting of the 2x custom machined channels.
In addition, some of the original standoffs and cross bars for supporting the heatsink would need to be replaced.
We can offer this conversion kit at cost, at 60 Euros per case including Paypal and registered post.
The cost of the SK466s would be additional.
At Buerklin, a standard length of 50mm costs 6 Euros incl VAT at quantities of 100.
So total cost is around 90~100 Euros.
https://www.buerklin.com/default.as...yp-Fischer-SK-466&event=ShowDvNr(B114400)&l=d
Thermal solutions are unfortunately not cheap.
We cannot use a standard U-channel because they simply do not have enough thickness for proper heat transfer.
Other simpler (lower cost) solutions we tried always exceed 40°C temperature rise.
You might perhaps use the CRC first and wait till the test team to report on subjective impressions before deciding.
Patrick
Probably best done by someone living in Germany, Holland, or Austria, preferably with a company.
There are two through holes required to bolt them onto the U-Channel.
If Fischer charges too much for the drilling, there is no reason you cannot drill yourselves.
Then you would need a conversion kit from us, consisting of the 2x custom machined channels.
In addition, some of the original standoffs and cross bars for supporting the heatsink would need to be replaced.
We can offer this conversion kit at cost, at 60 Euros per case including Paypal and registered post.
The cost of the SK466s would be additional.
At Buerklin, a standard length of 50mm costs 6 Euros incl VAT at quantities of 100.
So total cost is around 90~100 Euros.
https://www.buerklin.com/default.as...yp-Fischer-SK-466&event=ShowDvNr(B114400)&l=d
Thermal solutions are unfortunately not cheap.
We cannot use a standard U-channel because they simply do not have enough thickness for proper heat transfer.
Other simpler (lower cost) solutions we tried always exceed 40°C temperature rise.
You might perhaps use the CRC first and wait till the test team to report on subjective impressions before deciding.
Patrick
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.