Some not so good news, and this relates to the power supply heat sinks (regulators and rectifier diodes).
In the original design, we were intending to use a modified Fischer Elektronik (Germany) SK88-134 heat sink, which has a rated Rth of 0.75 °C/W. The total dissipation per channel of rectifier + regulator was expected to be 32W. Even though we need to shorten the height from 50mm to 40mm, we still expect temperatures below well 60°C. So that was fine.
Because the SK88 extrusions are physically heavy and thus costly to ship from Germany to Hong Kong for machining, and probably even more expensive for getting them machined in Germany, we sourced locally a replacement sink of comparable dimensions with the assumption that the thermal performance would be similar. And if you can recall, we made them available for 10 Euros a piece including machining and anodising. So you are paying 20 Euros for a pair per case
Finally in the last two weeks, Mark got all parts working and had a chance to test the power supply heat sink in circuit. And he got over 80°C after 30 minutes, which is too high even for us. There are a number of reasons for this. Firstly the regulator Mosfets has about 20% higher actual Vgs than we assumed from the data sheet. This increased the dissipation to 36W. The actual Rth of the heat sink then turned out to be about 1.8 °C/W, quite some way off the 1°C/W we were assuming. We take 100% responsibility for this, as we should have tested earlier. And I apologise openly for my sloppiness in publishing solutions without testing. We just did not have the time to test everything while we know quite a few people were waiting impatiently.
As an interrim solution, I have proposed at the F5X thread to use 12x Fukushima MPC74 1R 5W bulk film ceramic resistors per channel to run the power supply as CRC (passive). This results in the same output voltage as the regulator, but with a bit less ripple filtering. We expect this to be a totally viable solution with very little performance penalties, if any at all. We already published a PCB layout which is 1:1 compatible to the regulator PCB, so it is easy to implement in the same case layout without any changes.. Because the MPC74 are free standing power resistors with full rating even at 60°C ambient (and we are only using them at l1.8W each), they do not need to be thermally coupled to the power supply heat sinks. This means the power supply heat sinks are now only there for the rectifier diodes, and they dissipate 14W per channel instead. This comes to a temperature rise of less than 30°C. So the thermal world is in order again.
The mains relay has to be configured slightly differently and will now only turn on after pressing the pushbutton once (stand-by mode). This can be done easily with the protection board with one jumper (already planned).
To make up for our mistakes and to make thing simple, we shall make available a set of PCBs & MPC74's for 2 channels of CRC solution per case at out cost. So you can start building the amp without delay, while we test out further solution ideas for the regulator. You need to group the resistors into 4 groups of 6 (2 series group of 3 in parallel within the 6 to give 0R67) so that they give about the same resistance in the end. Some minor differences are not going to be important, as the rails do not have to be 100.00% symmetrical and identical.
We feel comfortable enough that we have offered you a good solution at minimum cost.
Patrick
In the original design, we were intending to use a modified Fischer Elektronik (Germany) SK88-134 heat sink, which has a rated Rth of 0.75 °C/W. The total dissipation per channel of rectifier + regulator was expected to be 32W. Even though we need to shorten the height from 50mm to 40mm, we still expect temperatures below well 60°C. So that was fine.
Because the SK88 extrusions are physically heavy and thus costly to ship from Germany to Hong Kong for machining, and probably even more expensive for getting them machined in Germany, we sourced locally a replacement sink of comparable dimensions with the assumption that the thermal performance would be similar. And if you can recall, we made them available for 10 Euros a piece including machining and anodising. So you are paying 20 Euros for a pair per case
Finally in the last two weeks, Mark got all parts working and had a chance to test the power supply heat sink in circuit. And he got over 80°C after 30 minutes, which is too high even for us. There are a number of reasons for this. Firstly the regulator Mosfets has about 20% higher actual Vgs than we assumed from the data sheet. This increased the dissipation to 36W. The actual Rth of the heat sink then turned out to be about 1.8 °C/W, quite some way off the 1°C/W we were assuming. We take 100% responsibility for this, as we should have tested earlier. And I apologise openly for my sloppiness in publishing solutions without testing. We just did not have the time to test everything while we know quite a few people were waiting impatiently.
As an interrim solution, I have proposed at the F5X thread to use 12x Fukushima MPC74 1R 5W bulk film ceramic resistors per channel to run the power supply as CRC (passive). This results in the same output voltage as the regulator, but with a bit less ripple filtering. We expect this to be a totally viable solution with very little performance penalties, if any at all. We already published a PCB layout which is 1:1 compatible to the regulator PCB, so it is easy to implement in the same case layout without any changes.. Because the MPC74 are free standing power resistors with full rating even at 60°C ambient (and we are only using them at l1.8W each), they do not need to be thermally coupled to the power supply heat sinks. This means the power supply heat sinks are now only there for the rectifier diodes, and they dissipate 14W per channel instead. This comes to a temperature rise of less than 30°C. So the thermal world is in order again.
The mains relay has to be configured slightly differently and will now only turn on after pressing the pushbutton once (stand-by mode). This can be done easily with the protection board with one jumper (already planned).
To make up for our mistakes and to make thing simple, we shall make available a set of PCBs & MPC74's for 2 channels of CRC solution per case at out cost. So you can start building the amp without delay, while we test out further solution ideas for the regulator. You need to group the resistors into 4 groups of 6 (2 series group of 3 in parallel within the 6 to give 0R67) so that they give about the same resistance in the end. Some minor differences are not going to be important, as the rails do not have to be 100.00% symmetrical and identical.
We feel comfortable enough that we have offered you a good solution at minimum cost.
Patrick
Last edited:
For the longer term, we are testing a number of possible solutions experimentally, all at our costs, to see what improvement we can get at full regulator thermal load. There are larger heatsink profiles from Fischer such as SK98-134, which will fit in the case. But they need substantially machining, which might be expensive at Fischer and costly to transport to Mark. The other easy solution is to use fans, which easily reduces the temperature by half, but it is probably one you don’t like.
So we went back to basics for maximum volumetric efficiency for sinks. And that means vertical, short fins on both sides of the regulator. The currently favored solution is 2 pieces of Fischer SK466-38, bolted on each side to a custom machined U-channel, which holds the regulator PCB and rectifier diodes, provides thermal interface to the sinks, and mounting interface to the cradle. Machining of the Fischer SK466-38 extrusions is then reduced to cutting to length and drilling 4x 3.2mm holes. You would need 4 SK466’s per case, and I expect them to cost under 10 Euros a piece including VAT, for a group purchase of 100 pieces directly from Fischer.
Once we have firm experimental results, we shall publish them, and then you can decide which one suits you best. We shall provide drawings for Fischer and machining aids for a conversion kit as necessary. I expect some of you to be happy with the CRC solution and not bothered to change back to regulator at all. But we’ll see.
What can you then do with the Regulator Mosfets if you go CRC ? They are ideal for trying out one of the configurations the test team is going through right now, where the 2SK3487 / 2SJ618 are used to replace the 2SK1530 / 2SJ201 as amplifier output devices. That particular prototype is already working, but we want to do all the fine tuning, test a few more options (with or without cascode, etc.), and run them through listening tests, before we publish the results all in one go. And if you really don’t want to test them at all, drop us a line and we are happy to take them back from you, as long as you got them from us in the first place via the Regulator Mosfet GB about 1 year ago. Or you can sell them at DIYA as matched pairs. I am sure a lot of F5X builder would be very interested at the price you got them for.
Patrick
So we went back to basics for maximum volumetric efficiency for sinks. And that means vertical, short fins on both sides of the regulator. The currently favored solution is 2 pieces of Fischer SK466-38, bolted on each side to a custom machined U-channel, which holds the regulator PCB and rectifier diodes, provides thermal interface to the sinks, and mounting interface to the cradle. Machining of the Fischer SK466-38 extrusions is then reduced to cutting to length and drilling 4x 3.2mm holes. You would need 4 SK466’s per case, and I expect them to cost under 10 Euros a piece including VAT, for a group purchase of 100 pieces directly from Fischer.
Once we have firm experimental results, we shall publish them, and then you can decide which one suits you best. We shall provide drawings for Fischer and machining aids for a conversion kit as necessary. I expect some of you to be happy with the CRC solution and not bothered to change back to regulator at all. But we’ll see.
What can you then do with the Regulator Mosfets if you go CRC ? They are ideal for trying out one of the configurations the test team is going through right now, where the 2SK3487 / 2SJ618 are used to replace the 2SK1530 / 2SJ201 as amplifier output devices. That particular prototype is already working, but we want to do all the fine tuning, test a few more options (with or without cascode, etc.), and run them through listening tests, before we publish the results all in one go. And if you really don’t want to test them at all, drop us a line and we are happy to take them back from you, as long as you got them from us in the first place via the Regulator Mosfet GB about 1 year ago. Or you can sell them at DIYA as matched pairs. I am sure a lot of F5X builder would be very interested at the price you got them for.
Patrick
Last edited:
Last but not least a few words about possible 2nd batch and preamp case.
As you realise, I have put Mark and the factory under a lot of stress because of various cosmetic issues, even though the parts themslves are fully to tolerance and functional specs. Both need time to recover from this ordeal, so we shall not consider any further batches for the next 6 months.
Also considering the complications (both logistics and delivery) with Conrad, we have decided that we shall only do a second batch if we can souce our own heatsinks locally at an acceptable quality and price.
I guess you and the test team would have enough DIY to do for the next 6 months anyhow, but I thought I should mention this upfront just to make sure there is no false expectations.
And all credits to Mark for getting this through.
Patrick
As you realise, I have put Mark and the factory under a lot of stress because of various cosmetic issues, even though the parts themslves are fully to tolerance and functional specs. Both need time to recover from this ordeal, so we shall not consider any further batches for the next 6 months.
Also considering the complications (both logistics and delivery) with Conrad, we have decided that we shall only do a second batch if we can souce our own heatsinks locally at an acceptable quality and price.
I guess you and the test team would have enough DIY to do for the next 6 months anyhow, but I thought I should mention this upfront just to make sure there is no false expectations.
And all credits to Mark for getting this through.
Patrick
Many other team members have made major contributions to the project. Above all Mark Lai has shouldered most of the donkey work in arranging all mechanical parts. Dave (fitzfish) and Alexis (ashaw) have both put in hundred of hours perfecting the PCBs. John, Greg, Fred, Nic & the regional distributors have arranged various GBs, put in hours and hours to do matching, .... to only name a few. I am the lucky one who only picked the interesting stuff to do. So credit & thanks are due to them all. And my apologies if I have missed any names.
We were also hit with bad luck this year. Apart from being occupied with the daily job, 4 members of the team have at various stages had family emergencies to take care of that took them out of actions for months. My thanks go especially to them, despite those difficulties.
Please don't forget to decide whether you (collectively) want us to air parcel one set parts to John for heatsink fitting.
Patrick
We were also hit with bad luck this year. Apart from being occupied with the daily job, 4 members of the team have at various stages had family emergencies to take care of that took them out of actions for months. My thanks go especially to them, despite those difficulties.
Please don't forget to decide whether you (collectively) want us to air parcel one set parts to John for heatsink fitting.
Patrick
Last edited:
Collective decisions are slow and we cannot wait weeks to take this one. I think a final check of the heat sinks is a good idea and worth the 3 euro/case. I assume that most will agree and I will therefore cover the expense for case subscribers that post here that they do not want, or are unable, to contribute to this "optional".
I will not wait for the collective approval of my suggestion
Cheers,
Nic
Yes, I want actual component checking. These chassis are high value being posted/carriered around the world.
There is little to be gained by taking the risk that one or a few holes are misplaced by a few tenths of a millimetre.
I could possibly alter modify if the error suits. Many of our Members need first time success/accuracy.
There is little to be gained by taking the risk that one or a few holes are misplaced by a few tenths of a millimetre.
I could possibly alter modify if the error suits. Many of our Members need first time success/accuracy.
Thanks Patrick for so much effort, I think I can say we are all in total admiration.
NicMac your offer is very generous, hopefully 3 euro won’t break to many subscribers banks. I will include this as a “good will” payment once I get final costs from patrick. I’m sure you’ll see most if not all your money back.
I have some further good news, Conrad have completed setting up the gigs and have finished the first pair of heat sinks for fitting.
NicMac your offer is very generous, hopefully 3 euro won’t break to many subscribers banks. I will include this as a “good will” payment once I get final costs from patrick. I’m sure you’ll see most if not all your money back.
I have some further good news, Conrad have completed setting up the gigs and have finished the first pair of heat sinks for fitting.
1 set for 1 case = 2 PCBs + 24 unmatched MPC74 1R resistors
NicMac__2 sets PCB+R
johnwmclean__1 set PCB
pchw__1 set PCB+R
Nikon1975__1 set PCB+R
JBdV__2 sets PCB+R
Horio__2 sets PCB+R
Mravinsky__1 set PCB+R
APK__1 set PCB
JBL4435___1 set PCB (waiting list)
pmchoong_3 sets PCB+R
ramallo___1 set PCB
oneoclock___1 set PCB
heinz1___1 set PCB
Triode4___1 set PCB+R
alazira___1 set PCB+R
studley__1 set PCB+R
smyslow__1 set PCB+R
ashaw__8 set PCB+R
scranton_1 set PCB+R
29/19/30 (PCB/R-Kit/All Cases)
NicMac__2 sets PCB+R
johnwmclean__1 set PCB
pchw__1 set PCB+R
Nikon1975__1 set PCB+R
JBdV__2 sets PCB+R
Horio__2 sets PCB+R
Mravinsky__1 set PCB+R
APK__1 set PCB
JBL4435___1 set PCB (waiting list)
pmchoong_3 sets PCB+R
ramallo___1 set PCB
oneoclock___1 set PCB
heinz1___1 set PCB
Triode4___1 set PCB+R
alazira___1 set PCB+R
studley__1 set PCB+R
smyslow__1 set PCB+R
ashaw__8 set PCB+R
scranton_1 set PCB+R
29/19/30 (PCB/R-Kit/All Cases)
Thank you very much, Patrick, for your status report.
As almost always, the old idiom "The devil is in the detail(s)" holds true. Mr. Murphy and his law might also be held responsible. In another context, you mentioned the German proverb "Der Weg ist das Ziel". But now, you are making a detour and I hope that this detour will not take very long and distract you from the other fine things you want to accomplish.
John, you can be sure that you will receive the "good will" payment for the surplus transportation costs from me, too.
I shall also sign up for a set of resistors:
1 set for 1 case = 2 PCBs + 24 unmatched MPC74 1R resistors
NicMac__2 sets PCB+R
johnwmclean__1 set PCB
pchw__1 set PCB+R
Nikon1975__1 set PCB+R
JBdV__2 sets PCB+R
Horio__2 sets PCB+R
Mravinsky__1 set PCB+R
APK__1 set PCB
JBL4435___1 set PCB (waiting list)
pmchoong_3 sets PCB+R
ramallo___1 set PCB
oneoclock___1 set PCB
heinz1___1 set PCB+R
Triode4___1 set PCB+R
alazira___1 set PCB+R
studley__1 set PCB+R
smyslow__1 set PCB+R
ashaw__8 set PCB+R
scranton_1 set PCB+R
CanAm Man__1 set PCB+R
29/20/30 (PCB/R-Kit/All Cases)
As almost always, the old idiom "The devil is in the detail(s)" holds true. Mr. Murphy and his law might also be held responsible. In another context, you mentioned the German proverb "Der Weg ist das Ziel". But now, you are making a detour and I hope that this detour will not take very long and distract you from the other fine things you want to accomplish.
John, you can be sure that you will receive the "good will" payment for the surplus transportation costs from me, too.
I shall also sign up for a set of resistors:
1 set for 1 case = 2 PCBs + 24 unmatched MPC74 1R resistors
NicMac__2 sets PCB+R
johnwmclean__1 set PCB
pchw__1 set PCB+R
Nikon1975__1 set PCB+R
JBdV__2 sets PCB+R
Horio__2 sets PCB+R
Mravinsky__1 set PCB+R
APK__1 set PCB
JBL4435___1 set PCB (waiting list)
pmchoong_3 sets PCB+R
ramallo___1 set PCB
oneoclock___1 set PCB
heinz1___1 set PCB+R
Triode4___1 set PCB+R
alazira___1 set PCB+R
studley__1 set PCB+R
smyslow__1 set PCB+R
ashaw__8 set PCB+R
scranton_1 set PCB+R
CanAm Man__1 set PCB+R
29/20/30 (PCB/R-Kit/All Cases)
Since I am ordering CRC PCBs anyhow, I might as well order the rest for the gang.
So 1 set of PCB per case will include :
2x CRC PCB (paid by us, free)
2x F5X Amp Beta boards (double sided 70µm, blue masks both sides)
8x PSU Capacitor boards (single sided 70µm, blue masks both sides)
1x Protection & Control board (double sided 35µm, blue masks both sides)
Estimated cost about 75 Euros plus post plus Paypal.
I presume you all want one set (for the case GB subscribers only, limited to 1 set per case).
The protection board is still awaiting testing and modification by Alexis, who is temporary out of action due to family matters. As soon as he gives the go, I can organise them accordingly.
Patrick
So 1 set of PCB per case will include :
2x CRC PCB (paid by us, free)
2x F5X Amp Beta boards (double sided 70µm, blue masks both sides)
8x PSU Capacitor boards (single sided 70µm, blue masks both sides)
1x Protection & Control board (double sided 35µm, blue masks both sides)
Estimated cost about 75 Euros plus post plus Paypal.
I presume you all want one set (for the case GB subscribers only, limited to 1 set per case).
The protection board is still awaiting testing and modification by Alexis, who is temporary out of action due to family matters. As soon as he gives the go, I can organise them accordingly.
Patrick
Hi Nic and JBL4435,
We've reserved 2 sets out of the 28 set GB. Now the 2 extra sets go to you.
However, the Front panel, back bracket & back panel are blank for these 2 reserved kit.
Blank Front panel means only pocket for button only. Back panel & bracket only has mounting holes and not holes for IEC, output terminals etc.
If 2 of you want option 2, I can modify all of them. BUT the edges (milled holes) will not be anodized as all parts are anodized already. I believe it's OK since you won't see the holes edge after the connectors are installed.
After all, I've to ask for your agreement before I go ahead.
Mark
We've reserved 2 sets out of the 28 set GB. Now the 2 extra sets go to you.
However, the Front panel, back bracket & back panel are blank for these 2 reserved kit.
Blank Front panel means only pocket for button only. Back panel & bracket only has mounting holes and not holes for IEC, output terminals etc.
If 2 of you want option 2, I can modify all of them. BUT the edges (milled holes) will not be anodized as all parts are anodized already. I believe it's OK since you won't see the holes edge after the connectors are installed.
After all, I've to ask for your agreement before I go ahead.
Mark
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.