and the difference is?
if you are thinking of chassis ambient is much higher, then think about what that ambient will be in a cold room vs in a hot room.
Lets analyse this scenario. - assuming your amp is working properly, in a cold room, your sink temp is still very high (your output devices is still working!!) the lower the outside temp, the biger the temp difference between the sink and ambient, faster the rate of heat loss. The temp of your sink will drop only if the rate of heat loss >> the heat genereated by the devices.
I agree abolutely that higher the sink temperature, faster the heat loss - that's assuming that temp diff between the sink and its surrounding is getting bigger.
Again, imagine you place the working amp in a room at say 100oC, your heat sink may be at 80oC, what happen then??
Regards
how the heck can the sinks be 80c when the ambient is 100c?
your sinks will be xc above ambient. say 25c above ambient. at 20c ambient, sinks are 45c. at 100c ambient, sinks are 125c.
PS: those are not exact, becouse of the higher effectivity of the sinks at 100c ambient.
this is why ex Conrad heatsinks is measured at 80c ambient.
your sinks will be xc above ambient. say 25c above ambient. at 20c ambient, sinks are 45c. at 100c ambient, sinks are 125c.
PS: those are not exact, becouse of the higher effectivity of the sinks at 100c ambient.
this is why ex Conrad heatsinks is measured at 80c ambient.
Last edited:
My bad for not making it clear. I was rushing to work this morning.
If the sink temperature is at say 60oC in a room at 25oC after reaching equilibrium, then you move the amp quickly to another room at 60oC. One will see the temperature of the sink quickly rise above 60oC. Why, heat loss is from the sink is severely impacted by the temperature difference (the lack of) wrt the environment. On the other hand, moving the amp to a room at 0oC will render the sink temperature drop below 60oC. The lower room temperature facilitate the heat loss of the sink.
My point - how effective is your heatsink in heat dissipation, which resulted in temperature change depends on the temperature difference between the sink and its environment, other things being equal.
One thing for sure, nobody like high sink temperature, does not matter how effective the heat dissipation is.
Regards,
If the sink temperature is at say 60oC in a room at 25oC after reaching equilibrium, then you move the amp quickly to another room at 60oC. One will see the temperature of the sink quickly rise above 60oC. Why, heat loss is from the sink is severely impacted by the temperature difference (the lack of) wrt the environment. On the other hand, moving the amp to a room at 0oC will render the sink temperature drop below 60oC. The lower room temperature facilitate the heat loss of the sink.
My point - how effective is your heatsink in heat dissipation, which resulted in temperature change depends on the temperature difference between the sink and its environment, other things being equal.
One thing for sure, nobody like high sink temperature, does not matter how effective the heat dissipation is.
Regards,
Lotse has a better understanding than you appear judging by the content quoted above.
The DIFFERENCE in Temperature is what controls the C/W.
You seem unable to grasp that.
Which I hope is enough, since in my case peak A would be 6A with a 3A bias while still in Class A operation. And If I'm not mistaken I wouldn't want the diode to start conducting until after that.
However, after looking at the MUR3020W graph, I'm having a hard time predicting at what temperature it starts to conduct in relation to voltage. They only give Tj values of 25,150,175 C. How would you relate junction temp to case temperature, and then to heatsink temperature.
I can get a thermistor in the heatsink and diode case, but Tj?
The question now is, at a heatsink temp of 25C above ambient, what is the junction temperature of the diode, and at what voltage will it start conducting.
From the delta T between the sink and Tj we can at least reasonably predict whether or not the main heatsink is enough to prevent runaway conditions.
When the diode is not conducting and the temperature stable, it's safe to assume that the junction temperature is the same as the heat sink temperature. Once the diode starts conducting, to calculate junction temperature rise above the sink temperurature multiply the power dissipation by (Rjc + Rti) where Rti is the insulator thermal impedance. Assume you're using keratherm and MUR3020s from the data sheets it's Pd * (1.5 +.1). At 2W average, the rise is 3.2 degrees, or junction temperature is 28.2C on 25 C sinks.
From the MUR3020 data sheet junction to air is 40C/W, so at 2 W average inside an amp at 30C without sinks the diodes' junctions will be at 110C, not terribly hot. As long as your source resistors are low enough to keep the diodes from conducting until you hit high power, it seems you don't really need to heat sink the diodes unless you intend to really push your amp.
From the MUR3020 data sheet junction to air is 40C/W, so at 2 W average inside an amp at 30C without sinks the diodes' junctions will be at 110C, not terribly hot. As long as your source resistors are low enough to keep the diodes from conducting until you hit high power, it seems you don't really need to heat sink the diodes unless you intend to really push your amp.
Thanks Bob! Just what I needed.
Just a quick question though. You're assuming the heatsink is 30C. Shouldn't it be closer to 50C, assuming ambient is 25C plus 25C rise due to the power transistors? So the junction temperature would be 80C + 50C + 3.2C = 133.2C?
Which now means that as per design the diode shouldn't conduct until at least 0.4V or so from the graph and I'm pretty much ok to mount on the heatsink...
Or, if I'm really afraid, or have the need to bias even higher... (hey why not) and I use the little heatsinks that I mentioned originally with only a 6.3C/W dissipation and assuming its about 5 C lower within the case than the heat sink temp. I would get. 80C - 12.6C +50C -5C +3.2C = 115.6C at the junction?
Is my assumption and math correct?
Last edited:
McBob gave two examples.
a.) Diode mounted on the main heatsink.
b.) Diode in free (inside) air
For the "on heatsink" the deltaT is 3.2C
For the "free air" the deltaT is 80C
There is a third option.
c.) Diode on separate heatsink
Assume a 20C/W sink and 3C/W Rth j-s Then @ 2W average, deltaT ~46C
All of these heat models depend very much on how you decide to set up your Rs & Diode voltages.
d.) The original F5 using 1.3A bias across a 0r47 gives 611mV across the Diode. The diode will get very hot and the amp will almost certainly be unstable.
e.) Reducing the bias per device to 1A across a 0r47 gives 470mV across the diode. The diode will be cool (low deltaT) when no output power is delivered, but will quickly warm/heat up even when the amplifier stays in ClassA. It was for that reason that a long while back, I raised the issue of "deciding the bias set up" you require.
The third and my preferred is a low Rs value.
f.) 1A bias & Rs<=0r3 giving <300mV across the diode in quiescent condition. = zero deltaT. Rising to <600mV for currents up to the ClassA limit. = short term transients rising to <600mV across the diode.
Only extreme peaks in ClassAB take the diode into high current mode. This short term heating has a low average heating value and the diodes stay cool to warm. The amplifier remains stable and the diodes only require a small heatsink.
There may be a benefit in using the main heatsink during this mode, but I have not seen an explanation, nor any data, that shows what the benefit could be, nor how big that benefit could be.
Hi community
first, its a great threat for F5 and F5T amp building, and a great thx for mr. Pass to give us these great amps.
We are 3 enthusiast in Germany and we will build 3x F5T V1/V2 amp.
We have make several test's with the diyaudio.com F5 board's
We use the Toshiba Power Fets for the amp, its sound's a little bit better as the IRF types.
The Power Supplay have +/- 29V DC and 14 amp per rail.
Afte read the threat, I have a last question to you, what is the better solution for best sound and distortion for this amp?
a:2 parallel Power Transistor with 1,5 Amp Bias per unit,
b:3 parallel Power Transistor with 1Amp Bias per Unit.
In summary, we will have 3 amps.
The Headsink is 500x210x40mm per Chanel from the Galaxy2000 U5-500mm Case by www.Modushop.biz
kind regards
Olaf
Sorry for my bad englisch and thx for you help.
first, its a great threat for F5 and F5T amp building, and a great thx for mr. Pass to give us these great amps.
We are 3 enthusiast in Germany and we will build 3x F5T V1/V2 amp.
We have make several test's with the diyaudio.com F5 board's
We use the Toshiba Power Fets for the amp, its sound's a little bit better as the IRF types.
The Power Supplay have +/- 29V DC and 14 amp per rail.
Afte read the threat, I have a last question to you, what is the better solution for best sound and distortion for this amp?
a:2 parallel Power Transistor with 1,5 Amp Bias per unit,
b:3 parallel Power Transistor with 1Amp Bias per Unit.
In summary, we will have 3 amps.
The Headsink is 500x210x40mm per Chanel from the Galaxy2000 U5-500mm Case by www.Modushop.biz
kind regards
Olaf
Sorry for my bad englisch and thx for you help.
Yes, Dazed, I forgot to say "above" ambient when describing the heat sink temperature. So with the diodes mounted on the main sinks which sit 25 C above ambient in a 25C room, the junction temperature would be 53.2 C.
The second example, free air inside a case with an internal temperature of 30C the junction would be 110C. You'd have to measure your case's internal temperature where your diodes are, it depends on the airflow inside the case and could be significantly higher than 30C. This local temperature also feeds into AndrewT's third example of separate heat sinks for the diodes.
While the free air and separate heat sink options will give you lower junction temperatures, IMHO there are two advantages to mounting the diodes on the main sink.
When using the store F5T boards, it's convenient. In normal home use where average power is low thee diodes will rarely conduct and as AndrewT points out heating will be minimal even if they do. This gives a stable operating point for the diodes.
Here's yet another spot where I may be out of my depth
- If the amp is driven well into diode conduction, it stands to reason the the amp is going deeper into class AB. In that case, the heat sinks would tend to get cooler, reducing the current the diodes will want to pass at the given voltage. This seems like it would decrease the chance of thermal runaway, but the effect would be slow and slight. Anyone else have any thoughts on this? I don't intend to test this theory with my upcoming F5T, I am not that much of a F.A.B.
The second example, free air inside a case with an internal temperature of 30C the junction would be 110C. You'd have to measure your case's internal temperature where your diodes are, it depends on the airflow inside the case and could be significantly higher than 30C. This local temperature also feeds into AndrewT's third example of separate heat sinks for the diodes.
While the free air and separate heat sink options will give you lower junction temperatures, IMHO there are two advantages to mounting the diodes on the main sink.
When using the store F5T boards, it's convenient. In normal home use where average power is low thee diodes will rarely conduct and as AndrewT points out heating will be minimal even if they do. This gives a stable operating point for the diodes.
Here's yet another spot where I may be out of my depth
- If the amp is driven well into diode conduction, it stands to reason the the amp is going deeper into class AB. In that case, the heat sinks would tend to get cooler, reducing the current the diodes will want to pass at the given voltage. This seems like it would decrease the chance of thermal runaway, but the effect would be slow and slight. Anyone else have any thoughts on this? I don't intend to test this theory with my upcoming F5T, I am not that much of a F.A.B.
Olaf, 6L6 did some measurements recently that should answer your question. If I remember correctly, running the output mosfets harder produces lower distortion. That means that your option A 1.5A bias would sound better. See: http://www.diyaudio.com/forums/pass-labs/235191-some-observations-bias-vs-distortion.html
You have to look hard at your interface with the heat sinks, though. Very low thermal impedance isolators like Keratherm will be required to keep your junctions cool enough for reliable operation in your chosen case.
You have to look hard at your interface with the heat sinks, though. Very low thermal impedance isolators like Keratherm will be required to keep your junctions cool enough for reliable operation in your chosen case.
Highest rail voltage to use in a F5V3
I have read in this forum about the experience of members who have built their F5V3's with rail voltages of around 32V and 45V.
What do you think is the highest rail voltage that one can use with the cascoded V3 design posted by Nelson Pass?
I am thinking of using 34V secondary transformers in my F5V3 build which should yield around 45V rails loaded but wondering why not use higher voltage secondaries since I have not bought the transformers yet. I realize that Class A output will be dependent on the bias current and total AB output on the rail voltage.
Is there any disadvantage (sound quality or otherwise) in using the highest rail voltage possible in the V3 design? Could the board design prove to be a limiting factor as to the voltage used?
Thanks.
Nash
I have read in this forum about the experience of members who have built their F5V3's with rail voltages of around 32V and 45V.
What do you think is the highest rail voltage that one can use with the cascoded V3 design posted by Nelson Pass?
I am thinking of using 34V secondary transformers in my F5V3 build which should yield around 45V rails loaded but wondering why not use higher voltage secondaries since I have not bought the transformers yet. I realize that Class A output will be dependent on the bias current and total AB output on the rail voltage.
Is there any disadvantage (sound quality or otherwise) in using the highest rail voltage possible in the V3 design? Could the board design prove to be a limiting factor as to the voltage used?
Thanks.
Nash
No............Here's yet another spot where I may be out of my depth
While the amplifier is in ClassA the total power sent by the PSU to the amplifier is constant.
That total power is shared between the load and the amplifier devices.
At quiescent all of the total power is dissipated as heat.
At maximum output (25W for an F5) the speaker load gets 25W and the amplifier dissipates ~40W. This is the coolest that the output stages gets to.
If the amplifier goes into ClassAB for output currents exceeding the ClassA limit, i.e. Ipk > 2*bias the amplifier gets hotter again because the dissipation increases as output power increases beyond the ClassA limit.
The F5TurboV3 schematic only shows +V and -V while the F5V2 shows +32out rail voltge/-32V.
In the accompanying text Papa notes " Power Mosfets are commonly available to 200 volt ratings, which sets the next voltage limitation at +/- 100 V rails. This could give us 600 watts rms into 8 ohms. Probably best not to try that with this particular circuit".
There is no other mention made in the text about rail voltage.
Since capacitor voltage ratings are usually 50V and then 63V I figure why not rails around 55V especially since there is no significant cost penalty for doing it as opposed to 45V rails.
The cost penalty is heat, less current through each devise and probably more devises.
Rush
Yes, but more output devices makes it easier to drive current and (in my opinion) sounds better.
The reduction in distortion as you raise bias is mainly lowering the 2nd harmonic. You are able, in an F5T, to adjust your 2nd harmonic with P3 - so I would have as many output devices as you can fit, and adjust to your taste with P3.
nasbap said:
As others have mentioned, heat is going to be your main issue. If you have lots of heatsink, go for it. You may also have to trim the operating points of the Cascodes to keep the Jfets in their safe range.
Olaf S said:
My specific recommendation to you - use 3 devices and bias until the transistor package is 55C with heatsinks 65C. However, either configuration you make will sound fantastic.
Last edited: