Well, Power up and initial biasing went according to plan and is now sitting at just over an amp. I'll bump it up in steps to ~1.4A and see what the temps look like. I picked up a thermal camera last year to help locate faults and failing connections in marine DC systems and it came to mind during power up. The hottest point on board appears to be the 0R47 source resistors (red crosshairs) which easily pass the touch test. Time to hook up a test speaker and populate the second channel 
Just a quick note to indicate the efficiency of fan cooling. The F5m ran all night making music with cover on and no fan. The heatsink temperature in the morning was 54°C, and presumably the bias remained at 310mV/0.66A or thereabouts.
I then placed a pair of 80mm AC Infinity fans directly over the locations of the PCBs on the heatsink, drawing upward. The fans were set on Low (1500 rpm, I think). 2.5 hours later I measured heatsink temperature and bias. The temp was 38°C, and the bias was 248mV R and 237mV L, or 0.53A and 0.50A.
Clearly the F5m has a temperature-dependent bias, as does the ACA Mini.
The implications of this are that (1) there might be a need for seasonable adjustment of bias, and (2) I don’t know what fraction of the observed temperature reduction is due to improved heat transfer by the fan versus the concomitant bias drop in the F5m’s circuits as the temperature dropped.
In any case, it does seem the cooling provided by the fan will allow for a significant increase in bias. I won’t know how much until I bring it down to the bench (dining room table), and do the actual tests. But right now, it is too much fun listening to this amp to bother...
I then placed a pair of 80mm AC Infinity fans directly over the locations of the PCBs on the heatsink, drawing upward. The fans were set on Low (1500 rpm, I think). 2.5 hours later I measured heatsink temperature and bias. The temp was 38°C, and the bias was 248mV R and 237mV L, or 0.53A and 0.50A.
Clearly the F5m has a temperature-dependent bias, as does the ACA Mini.
The implications of this are that (1) there might be a need for seasonable adjustment of bias, and (2) I don’t know what fraction of the observed temperature reduction is due to improved heat transfer by the fan versus the concomitant bias drop in the F5m’s circuits as the temperature dropped.
In any case, it does seem the cooling provided by the fan will allow for a significant increase in bias. I won’t know how much until I bring it down to the bench (dining room table), and do the actual tests. But right now, it is too much fun listening to this amp to bother...
@von Ah Very nice. I've wondered how much sustained airflow it would take to make a substantial difference but haven't played with it. Last night, with temps stabilized, I opened an overhead door and let 40F air spill into the shop (15-18deg below ambient). After 5 minutes the bias had only swung 12-15mV which surprised me and seems to illustrate the difference between ambient temp and forced air cooling. I think I'll buy a nocturna or similar quiet fan to experiment with now.
Sevententh's, I'm not sure I get your point about the difference between ambient temp and forced air. I'd say both can be effective. I'd argue that for your experiment with outside air, it would have to be more sustained than 5 minutes to have much effect. That is, there is so much thermal inertia in the sinks, especially if they are large, that 5 minutes of cool ambient air would have minimal effect on the attached transistors, which is what you saw.
I do think that cooler ambient air can be just as effective as forced air. Consider the case of Pass's write up for the F5m, where for 3 specific chassis/heatsinks, he gives the max bias recommendation, a bias level that will produce a 30°C temp rise on the sinks. My understanding of this is that, assuming your room is at 20°C, a 30°C rise will get you to 50°C for the sinks, which translates to a safe level for the dies inside the transistor bodies (which are at a much higher temp, but must be kept below the max junction temperature rating of 150°C).
If you make your room cooler than 20°C, then you can exceed the max bias Pass recommends for that chassis, and your sinks will still be at or below 50°C and therefore your dies will still be safe. Conversely, if your room is hotter than 20°C (or, like me, use poor judgement and pick a chassis with wimpy sinks), you will have to use a lower bias or risk letting your dies get too close to or exceed their 150°C junction temperature rating. Or use forced air.
I do think that cooler ambient air can be just as effective as forced air. Consider the case of Pass's write up for the F5m, where for 3 specific chassis/heatsinks, he gives the max bias recommendation, a bias level that will produce a 30°C temp rise on the sinks. My understanding of this is that, assuming your room is at 20°C, a 30°C rise will get you to 50°C for the sinks, which translates to a safe level for the dies inside the transistor bodies (which are at a much higher temp, but must be kept below the max junction temperature rating of 150°C).
If you make your room cooler than 20°C, then you can exceed the max bias Pass recommends for that chassis, and your sinks will still be at or below 50°C and therefore your dies will still be safe. Conversely, if your room is hotter than 20°C (or, like me, use poor judgement and pick a chassis with wimpy sinks), you will have to use a lower bias or risk letting your dies get too close to or exceed their 150°C junction temperature rating. Or use forced air.
@northpaw you'll get no argument from me. The thermal mass of even large finned aluminum was the surprising part. I guess I expected faster heat loss given the differential. This morning I powered the amp up and it sat for a while before I realized that I had neglected to check the bias on a cold start up. I powered it down with the shop doors open and watched the temp slowly drop until grabbing a battery and a tiny (1.5"/40cm) pancake fan and set it a few inches away. Despite the fact that forced air heat exchange is effectively used all around us, it was an eye opener and a nice lesson to watch it play out. Cheers.
Lurking around this early morning (still dealing with jetlag from Japan), I came across this interesting thread by Mark Johnson that may be helpful in this regard. He talks specifically about an internal fan, but it can apply to a heatsink fan as well.
@Holzarbeiter too.
https://www.diyaudio.com/community/...-temperature-controlled-analog-no-pwm.336346/
Thanks,
Looks to be very informative article. I 'm sure it will answer the questions I still have.
Channel #2 setup yesterday went well. Mounted to a 6" section of heatsinks USA 10.080" profile with the std 24V supply and bias set
at ~1.3A.
I logged the offset voltage for just over an hour from a cold start and it looks good. I'll power up both channels today to dial them in and start thinking about a case. The hobby budget took a big hit yesterday (our adopted rescue Saint needs significant dental work) and that forces a decision at this point. The higher WAF and ease of a bought case or something put together in the shop that's likely to be a little rough around the edges but allows the purchase of an inexpensive scope? Try not to laugh when I post up pics of a toaster oven with heatsinks ... cheers
I logged the offset voltage for just over an hour from a cold start and it looks good. I'll power up both channels today to dial them in and start thinking about a case. The hobby budget took a big hit yesterday (our adopted rescue Saint needs significant dental work) and that forces a decision at this point. The higher WAF and ease of a bought case or something put together in the shop that's likely to be a little rough around the edges but allows the purchase of an inexpensive scope? Try not to laugh when I post up pics of a toaster oven with heatsinks ... cheers
I actually have a large toaster oven/air fryer in my shop and after staring at it for a minute, I imagine that left to look the part, foil lined tray, fry basket and discolored glass door intact, it might be perfect for some place. That place is not the house my wife lives in however, so if/when I fire up my forge again, I'll have a means of heat treating steel and a heat source can be useful for a lot of things. I also imagine the smell of fried food emanating from it when operting temp is reached. Moving on.
My on-hand materials provide a couple options. An industrial steel look. An industrial steel, punk-ish design complete with stop sign/yield sign panels or build a small bending brake to make a reasonably nice floor pan and lid along with figured maple, black walnut, jatoba or whatever else is on my rack .
My on-hand materials provide a couple options. An industrial steel look. An industrial steel, punk-ish design complete with stop sign/yield sign panels or build a small bending brake to make a reasonably nice floor pan and lid along with figured maple, black walnut, jatoba or whatever else is on my rack .
Yesterday I pulled the F5m from the system it’s been in the past few days, running at a low bias of 0.66A without fans, and re-biased it with cooling fans running. I placed 2 80mm fans on top of the cover, over the heatsink locations where the PCBs were mounted. Room was at 18°C.
I started by setting the bias initially at about 490mV on an already warm unit, closing up, and waiting for it to stabilize.
With fan speed on Low, bias settled at 535mv L and 555mV R, with sinks at 54°C. Average Bias = 1.15A
With fan set to Medium, bias settled at 515mV L and 540mV R, with sinks at 50°C. Average Bias = 1.13A
Since I’d prefer to run with the fans on low and also keep the sinks near 50°C, I reduced the bias, closed up, and waited again for it to settle.
With fan speed on Low, bias settled at 419mv L and 409mV R, with sinks at 48°C. Average Bias = 0.88A
So there is some significant improvement with the fans, but was hoping for more than I got.
Today, looking for how to get more, I revisited the top cover and sides of the Niles chassis. These have a rather large number cooling slots in them, and the fans are pulling air through these slots as well as up through the screened openings below the heatsink. All those slots had a good purpose with the innards of the original Niles, but with the F5m the only heat producers of any significance not on the heatsink are the two thermistors on the ± rails. So I taped over most of the slots, leaving just a few near the rear so a little bit of air would be pulled through the interior.
This made a significant difference. This morning I was able to settle the bias at 506mV R and 517mV L with the sinks at 51°C. Average bias = 1.09A.
It is interesting to compare this to the results I had several days ago, before I tried the fans:
— for a bias of 435mV with an open cover, sinks were at 55°C.
— for a much lower bias of 300mV and cover on, sinks were at 60°F.
So a considerable improvement, and I’ve hit my hoped-for target of being able to bias at 1A or so with this small chassis, with fans on low.
While I had it open again, I took some of the suggestions made by @Ben Mah (post #1055) to move some items around to minimize potential noise from the AC circuitry (see photo below):
— rotated the transformer 90° CCW
— moved the bridge rectifier to the other side of terminal strip (further from amp PCBs)
— moved the PS PCB a bit further from the amp PCBs
I also attached a temperature sensor to the end of the heatsink (left in the photo below). This is part of a thermal trigger for the fans. This way, I don’t have to remember to turn the fans on or off, as I have them powered external to the F5m chassis.
I started by setting the bias initially at about 490mV on an already warm unit, closing up, and waiting for it to stabilize.
With fan speed on Low, bias settled at 535mv L and 555mV R, with sinks at 54°C. Average Bias = 1.15A
With fan set to Medium, bias settled at 515mV L and 540mV R, with sinks at 50°C. Average Bias = 1.13A
Since I’d prefer to run with the fans on low and also keep the sinks near 50°C, I reduced the bias, closed up, and waited again for it to settle.
With fan speed on Low, bias settled at 419mv L and 409mV R, with sinks at 48°C. Average Bias = 0.88A
So there is some significant improvement with the fans, but was hoping for more than I got.
Today, looking for how to get more, I revisited the top cover and sides of the Niles chassis. These have a rather large number cooling slots in them, and the fans are pulling air through these slots as well as up through the screened openings below the heatsink. All those slots had a good purpose with the innards of the original Niles, but with the F5m the only heat producers of any significance not on the heatsink are the two thermistors on the ± rails. So I taped over most of the slots, leaving just a few near the rear so a little bit of air would be pulled through the interior.
This made a significant difference. This morning I was able to settle the bias at 506mV R and 517mV L with the sinks at 51°C. Average bias = 1.09A.
It is interesting to compare this to the results I had several days ago, before I tried the fans:
— for a bias of 435mV with an open cover, sinks were at 55°C.
— for a much lower bias of 300mV and cover on, sinks were at 60°F.
So a considerable improvement, and I’ve hit my hoped-for target of being able to bias at 1A or so with this small chassis, with fans on low.
While I had it open again, I took some of the suggestions made by @Ben Mah (post #1055) to move some items around to minimize potential noise from the AC circuitry (see photo below):
— rotated the transformer 90° CCW
— moved the bridge rectifier to the other side of terminal strip (further from amp PCBs)
— moved the PS PCB a bit further from the amp PCBs
I also attached a temperature sensor to the end of the heatsink (left in the photo below). This is part of a thermal trigger for the fans. This way, I don’t have to remember to turn the fans on or off, as I have them powered external to the F5m chassis.
Hi all, I'm building an F5M using F5 boards and parts I have on hand. I see a post where the IRFP140 and IRFP240 are noted as interchangeable. I also see a post citing an error in the F5M Essentials Kit’s BOM listing IRFP140 with IRFP9240. Anyone know if pairing IRFP140 with an IRFP9240 would be bad decision? 😊
I guess I'm hung up on what interchangeable means between the IRFP140 and IRFP240 …… does it mean combinations IRFP140/IRFP9140, IRFP240/IRFP9240, IRFP140/IRFP9240, or IRFP240/IRFP9140 are acceptable? The post citing BOM error has me thinking it means use only IRFP140/IRFP9140 or IRFP240/IRFP9240.
Data sheets clearly show different measurements, but as a person who does not want to order parts and even worse desolder, I want to ask.
Thanks in advance!
-Donn
I guess I'm hung up on what interchangeable means between the IRFP140 and IRFP240 …… does it mean combinations IRFP140/IRFP9140, IRFP240/IRFP9240, IRFP140/IRFP9240, or IRFP240/IRFP9140 are acceptable? The post citing BOM error has me thinking it means use only IRFP140/IRFP9140 or IRFP240/IRFP9240.
Data sheets clearly show different measurements, but as a person who does not want to order parts and even worse desolder, I want to ask.
Thanks in advance!
-Donn
northpaw, Did the changes reduce the noise?
For cooling, decreasing the resistance to air flow to the intake at the chassis would increase cooling. Try increasing the distance between the bottom of the chassis and the supporting surface. This would be especially helpful if the existing gap is small.
For cooling, decreasing the resistance to air flow to the intake at the chassis would increase cooling. Try increasing the distance between the bottom of the chassis and the supporting surface. This would be especially helpful if the existing gap is small.
I suggest you post this in the F5 Turbo thread. Be careful changing things in this circuit without knowing exactly what you are doing. For example, by paralleling JFETs you are affecting OLG and load line, which potentially should or could be compensated. Having P3 is nice as most JFETs are not matched with degeneration, and using P3 you can ensure equal currents (zero DC between P and N) using P3, even without a distortion analyzer. This does not guarantee lowest possible distortion, but pretty close. Also, this circuit should not be less transparent than an F4. The original was a truth machine, and even though the T has less feedback it should be pretty transparent given it’s low distortion and high bw. Saying this because it may indicate you should take a closer look at your load line and feeback network, as you suggest. The T thread should be full of people having thoughts on this.
Hi Ben,
I don't sense an improvement in the slight buzz from the speakers. It's still something I can hear with my ear against the speaker, but not by the time I'm a foot away. Given this level, it is essentially a non-issue for me, from my listening position at 8'+ away.
The fan-related noise has some wrinkles to it. The inherent fan noise is quite low. It is not zero, but I could not hear the fan from my listening position in the original configuration. But now that I have taped off the broad areas that were allowing air to be drawn into the chassis more evenly, I can hear more fan-related noise. As your comment presages, it isn't the fan itself, but the wind noise from the more constrained air flow pattern. I can't hear it with most music playing, I do like a dead silence from a system when I listen to music that has quiet passages and even silent moments, so it is a concern.
The Niles chassis has 1/2" tall feet, pretty decent, but I went ahead an added another 1" height with wood blocks under them to see, but it was without effect. I then made a make-shift 8"-tall chimney out of a cardboard box that I slipped over the fans, as I was curious if that would block any noise, but it does not (as a side note, after 15 min it did seem to allow the heatsink temp to drop almost 1°C; edit: or maybe it was because I added those wood blocks first?). So my next guess to the source of the noise is most likely the funneling of air through the screen that is sandwiched between the bottom of the heatsink and the large holes in the bottom of the chassis (see 2nd image in post #1052). The next time I have it open, I may loosen the screws in the bottom of the heatsink and slip the screen out to see if it makes a difference. Until then, it is but a minor annoyance.
I've now listened to the F5m for a couple of hours at 1.09A bias, as well as earlier at 0.66A bias. Regardless of bias, this amp provides remarkable detail compared to other amps I've had. I would also say the sound is lean, and I mean that in a good way. There are no exaggerations, such as the chestiness that is sometime heard in radio announcer's voices, or an untoward "lushness" in some recordings. It just sounds honest, natural. What I seem to notice between the lower- and higher-bias settings that I've listened to so far is that the quality of certain sounds, like piano notes, differs. I don't want to sound too flowery and overly subjective about it, but the best analogy I can provide is that the higher-bias sound has piano notes that are more rounded and smooth, like a plump, ripe grape would be. It's a beautiful sound.
I don't sense an improvement in the slight buzz from the speakers. It's still something I can hear with my ear against the speaker, but not by the time I'm a foot away. Given this level, it is essentially a non-issue for me, from my listening position at 8'+ away.
The fan-related noise has some wrinkles to it. The inherent fan noise is quite low. It is not zero, but I could not hear the fan from my listening position in the original configuration. But now that I have taped off the broad areas that were allowing air to be drawn into the chassis more evenly, I can hear more fan-related noise. As your comment presages, it isn't the fan itself, but the wind noise from the more constrained air flow pattern. I can't hear it with most music playing, I do like a dead silence from a system when I listen to music that has quiet passages and even silent moments, so it is a concern.
The Niles chassis has 1/2" tall feet, pretty decent, but I went ahead an added another 1" height with wood blocks under them to see, but it was without effect. I then made a make-shift 8"-tall chimney out of a cardboard box that I slipped over the fans, as I was curious if that would block any noise, but it does not (as a side note, after 15 min it did seem to allow the heatsink temp to drop almost 1°C; edit: or maybe it was because I added those wood blocks first?). So my next guess to the source of the noise is most likely the funneling of air through the screen that is sandwiched between the bottom of the heatsink and the large holes in the bottom of the chassis (see 2nd image in post #1052). The next time I have it open, I may loosen the screws in the bottom of the heatsink and slip the screen out to see if it makes a difference. Until then, it is but a minor annoyance.
I've now listened to the F5m for a couple of hours at 1.09A bias, as well as earlier at 0.66A bias. Regardless of bias, this amp provides remarkable detail compared to other amps I've had. I would also say the sound is lean, and I mean that in a good way. There are no exaggerations, such as the chestiness that is sometime heard in radio announcer's voices, or an untoward "lushness" in some recordings. It just sounds honest, natural. What I seem to notice between the lower- and higher-bias settings that I've listened to so far is that the quality of certain sounds, like piano notes, differs. I don't want to sound too flowery and overly subjective about it, but the best analogy I can provide is that the higher-bias sound has piano notes that are more rounded and smooth, like a plump, ripe grape would be. It's a beautiful sound.