Yes, good point. I did stray beyond 1A bias in the last iteration (up to 1.14A), so I was beyond the recommended range for the 10Ω thermistors I used.
Presently I am re-biasing to a lower value using only passive cooling with the top cover off (still not entirely happy with the fan noise situation; turns out I really do like dead silence). I am shorting the RCAs this time. I will be able to get to about 0.9A without the cover, and will run that way for at least a few weeks -- I have to get to the other side of Tax Day in the US before taking a shot at rewiring things to rotate the PS board, etc. But before starting the re-bias, I did take a few of William's suggestions, and repositioned the rectifier and secondary wiring so the latter is not parallel to the DC rail wiring anymore. Hopefully through my inexperience I have not introduced any new horriblenesses...
Monoblocks have several nice advantages. I mainly did it because one Class A channel is easier for me to transport than two. My chassis and PS's are substantial (able to run >3A bias at normal FW rail voltages). I won't be using the F5m kit PS board, it obviously only included a single anyhow. No love for the monoblock guys....
Maybe you do need the series resistors in the supply lines, instead of the thermistors, like the standard F5 has.
The resistors will have the same filtering of the supply ripple (with same bias current setting and same capacitors),
whether at start up or later on.
That's a good catch by Dennis. That block rectifier will probably need to be able to dissipate a few watts so that guy should be mounted to the floor with a thin layer of thermal paste or something.
I'm not a fan of bundling the center tap wires with the +/- from the rectifier. I would separate those elements myself.
I'm not a fan of bundling the center tap wires with the +/- from the rectifier. I would separate those elements myself.
You see it correctly. But you are right, it is getting quite warm. Earlier, I had it on the chassis floor with a little thermal compound, but since I wasn't getting warm, I thought it would be OK to have it free, but obviously it isn't; that heatsinking was more effective than I thought. I'll stick a heatsink on its base, or reattach it to the case for now. This will all be dealt with in a more permanent manner when I rewire things in a few weeks. Thanks for noticing.
Meanwhile, this amp sounds just beautiful.
It needs to get rid of it’s heat, especially in the long run, and especially since it is alone. Tighten to at least 1Nm, better 1.2. M4, and use split washer. Maybe it will even make less noise, and following Williams splitting advice may prove beneficial also in that respect.
I was thinking along those lines (though I was unaware of what the original F5 had). But I know nearly nothing about PS design, and that is why I asked. I appreciated what Ben Mah said earlier and the simulation he ran, which led to the suggestion of adding additional capacitance at the end of the existing CRC filter. I'm assuming that is the most desirable way to approach/modify this, but I don't really know.
We are having a lot of discussion about this noise issue, and I am learning a great deal from it and I greatly appreciate the suggestions so many have contributed. That is one of the truly great things about the DIYAudio community. But I also want to say that the level of noise is really minor (I can't hear it more than several inches from the speakers); I would have additional concern about it if it is resulting in other degradation. My ears tell me things are pretty damn good, though.
If the noise is that low, then it's ok for diy, where you don't have to impress anybody with measurements.
The original F5 PS board has 30mF / 0.12R @ 12W / 30mF for each rail.
Be sure the 0.12R is rated for adequate watts, it was four 0.47R @ 3W resistors in parallel for the F5.
The original F5 PS board has 30mF / 0.12R @ 12W / 30mF for each rail.
Be sure the 0.12R is rated for adequate watts, it was four 0.47R @ 3W resistors in parallel for the F5.
Last edited:
@rayma - In post #1114, Ben estimated the R of the thermistor in a warmed-up F5m CRC at 0.5Ω, quite a bit higher than the 0.12Ω in the original F5 PS. It seems if I were to replace the thermistor with 0.12Ω resistor, would it not give me more noise, not less?
As I've said, I know next to nothing about PS design, so there are many places I can go wrong (witness today my misstep with not heatsinking the rectifier). I can read a graph, though. Below is the T dependence of the EPCOS B57234S0100M000 thermistor (the one in Nelson's F5m writeup) from the datasheet. It is 10Ω at room temp, as expected. It reaches 0.5Ω, Ben's estimate, at about 150°C. I doubt it gets that hot in the F5m; I'd guess something maybe as high as the 60-70°C range from touching them (I thought they were a little hotter than the heatsink). If so, it would be something on the order of 2Ω when the F5m is running. If this is correct, it seems I would need a >2Ω resistor to reduce noise more than the thermistor is already doing, with a max of 10Ω, because from my experiment this morning, 10Ω is enough to fully quiet my PS.
I'm not suggesting that I should replace the thermistor with a 5Ω or 10Ω resistor, as we both already agree the noise level with the stock thermistors do not give me a bothersome level of noise in practical listening. So as you have said, its ok for DIY, and I thank you for that reassurance, which I take as meaning that there is no other reason to worry about the presence of the noise level I am hearing. I do hope to reduce it with some rearranging in a few weeks and maybe adding some capacitance, but I'm also pretty happy with the thing as it exists now. Thanks again for commenting and helping me. It is much apprecaited.
As I've said, I know next to nothing about PS design, so there are many places I can go wrong (witness today my misstep with not heatsinking the rectifier). I can read a graph, though. Below is the T dependence of the EPCOS B57234S0100M000 thermistor (the one in Nelson's F5m writeup) from the datasheet. It is 10Ω at room temp, as expected. It reaches 0.5Ω, Ben's estimate, at about 150°C. I doubt it gets that hot in the F5m; I'd guess something maybe as high as the 60-70°C range from touching them (I thought they were a little hotter than the heatsink). If so, it would be something on the order of 2Ω when the F5m is running. If this is correct, it seems I would need a >2Ω resistor to reduce noise more than the thermistor is already doing, with a max of 10Ω, because from my experiment this morning, 10Ω is enough to fully quiet my PS.
I'm not suggesting that I should replace the thermistor with a 5Ω or 10Ω resistor, as we both already agree the noise level with the stock thermistors do not give me a bothersome level of noise in practical listening. So as you have said, its ok for DIY, and I thank you for that reassurance, which I take as meaning that there is no other reason to worry about the presence of the noise level I am hearing. I do hope to reduce it with some rearranging in a few weeks and maybe adding some capacitance, but I'm also pretty happy with the thing as it exists now. Thanks again for commenting and helping me. It is much apprecaited.
If you did an AP distortion measurement, you'd see the 120Hz ripple and some spurious distortion artifacts from it way down in the noise.
Nothing to worry about.
Thermistor Resistance in the CRC filter - for my unit, with its quirks and flaws, shows 1.2V across the thermistor for 0.9A bias current. Since there are no differential equations or Green's functions to solve, I can actually do the calculation for resistance: 1.3Ω, which translates to about 95°C on the T-dependence plot from the datasheet.
The second part of the kit PS board is an RC filter. As issued, the R is variable NTC with a starting value of either 10 Ohms or 4.7 Ohms. The C is a pair of 4700 uF. If you want to lower the R, then it is certainly possible to maintain a similar RC factor (time constant) by increasing the value of the capacitance. Maybe try a pair 10,000 uF or 15,000 uF caps. That will keep the background 120Hz pretty silent.