I probably editted my post after you replied, I found these guys at digikey MBRF1045-E3/45 Vishay Semiconductor Diodes Division | MBRF1045-E3/45GI-ND | DigiKey
Even with the greater thermal resistance of the FP version, you think that small Heatsink from earlier would be suitable? And still use thermal compound like you would with a metal on metal contact?
The thermal resistance in total is all that matters: just add them up (K per Watt = deg. C per Watt), multiply by the power dissipation (per diode) and you have the temperature rise - above the amp internal temperature. The die temperature should be lower than 175 deg. C (with a good margin)!
So for the part linked above, you have 4K/W from the die to the case (call it 5K/W from die to heatsink). Now add the heatsink thermal resistance (choose a 15K/W heatsink, say) - total of 20K/W.
If the amp is 35 deg. C internally, and each diode burns 2W in the worst case, the uplift is 40K = 40 deg. C, and the die temperature will be 75 deg. C (OK).
Given that the diode will burn only about 1.3W, we are safe with a margin.
So for the part linked above, you have 4K/W from the die to the case (call it 5K/W from die to heatsink). Now add the heatsink thermal resistance (choose a 15K/W heatsink, say) - total of 20K/W.
If the amp is 35 deg. C internally, and each diode burns 2W in the worst case, the uplift is 40K = 40 deg. C, and the die temperature will be 75 deg. C (OK).
Given that the diode will burn only about 1.3W, we are safe with a margin.
So I did a quick prototype with what parts I have, using the huge heatsinks which were on hand already with some of the non isolated mbr1045 diodes.
Wired up like this:
Surprisingly I measured very very close to this. I used a heatsinked 50W 3 ohm resistor load to simulate fairly close to the 3.25A draw of the 845.
The only thing I'm wondering about is that my voltage drifted ever so slightly back and forth (according to my cheap multimeter), between 10.09V and 10.12V.. It didn't seem to want to settle down even after about 15 minutes. Is this possibly just my meter? Or is it normal for the voltage to drift a bit like this..
I have a better multimeter on order, so I'll try again when I get that.
Also, What is a safe range for filament voltage on the 845? Am I okay anywhere between 9.5 and 10.5V? I'll fine tune it in, but just kind of curious about margin.
Wired up like this:
Surprisingly I measured very very close to this. I used a heatsinked 50W 3 ohm resistor load to simulate fairly close to the 3.25A draw of the 845.
The only thing I'm wondering about is that my voltage drifted ever so slightly back and forth (according to my cheap multimeter), between 10.09V and 10.12V.. It didn't seem to want to settle down even after about 15 minutes. Is this possibly just my meter? Or is it normal for the voltage to drift a bit like this..
I have a better multimeter on order, so I'll try again when I get that.
Also, What is a safe range for filament voltage on the 845? Am I okay anywhere between 9.5 and 10.5V? I'll fine tune it in, but just kind of curious about margin.
It is common, but the filament of 845 will be more "steady" than a resistor. Keep in mind that If use a hum balance, the consumption will be higher than 3.25A.
Within 9.5A the emission is fine, although fine tuning will not be so hard. In any case, as the filament will be aging, the voltage drop will vary a little.
Gee, a whole 0.3% wobble! Perfectly normal. The mains varies a little with evey second that passes.
9.5V to 10.5V is allowable, but even this will shorten the life of an 845, unless these extremes are brief.
If you are contemplating unregulated filament supply, please check the mains tolerance in your location. Many places, like here in the UK, +/- 10% supply tolerance is allowed, and that is certainly not good enough, unless you don't care about DHT lifetime.
The rectifier pulses are not completely eliminated, either, and combined with the big electrolytic across the filament - shorting the (differential) music signal there, usually makes for sound quality even worse than ac. Plenty of views about this type of filament supply on this forum, try a search.
If you are contemplating unregulated filament supply, please check the mains tolerance in your location. Many places, like here in the UK, +/- 10% supply tolerance is allowed, and that is certainly not good enough, unless you don't care about DHT lifetime.
The rectifier pulses are not completely eliminated, either, and combined with the big electrolytic across the filament - shorting the (differential) music signal there, usually makes for sound quality even worse than ac. Plenty of views about this type of filament supply on this forum, try a search.
I had considered the fact that mains voltage can vary, but where I live voltage is a steady 120VAC, so I'm not concerned about that.
As for unregulated dc on the filaments, I guess I was taking a queue from Thomas at the Vinylsavior blog on one of his 211 builds
VinylSavor: 211
I know that current regulated is the preferred way to go, but I was under the impression that passively filtered DC would still be preferrable to regulated voltage DC based on some of the threads I've read on here on the subject.
Please look carefully at the value of choke the Thomas is using - and especially the fact that a second is added, without a final capacitor. Then calculate the impedance of the final choke at the lowest audio frequency.
Both of these features are essential to the performance of passive supply that Thomas uses - the performance will be very different without them. Raw rectified dc is otherwise the worst of both worlds, sad to say.
Problems with the sound of regulation are certain where Voltage regulation is used. This is no surprise - the high-current regulator's feedback sensing is across the filament, and will cause the regulator to try to cancel the naturally occurring differential music signal here.
Both of these features are essential to the performance of passive supply that Thomas uses - the performance will be very different without them. Raw rectified dc is otherwise the worst of both worlds, sad to say.
Problems with the sound of regulation are certain where Voltage regulation is used. This is no surprise - the high-current regulator's feedback sensing is across the filament, and will cause the regulator to try to cancel the naturally occurring differential music signal here.
They are rather large chokes which I was aware of, but was thinking only about ripple reduction.
Impedance of the final choke in Thomas' design at 20Hz is 12.5 ohm (did I do that right? XL= 2πfL)
This would be much higher than the output impedance of the filament supply I posted wouldn't it?
I'll do some reading, but I'm not sure I understand what effect the output impedance of the filament supply has.
I'll be honest, I really did plan to go with your current regulated boards, but I'm concerned about real estate in the amp.. And the heat they produce is something I wasn't sure I wanted to get into dealing with.. Maybe I should just suck it up and figure out how to fit them into the design.. lol
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.