Temp of drivers in low impedance amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm getting ready to build a mini sound system. It's being built for huge output power, but small space, and I need to utilize my heatsinks space.

I want the 2 subwoofer amp channels to be capable of driving impedance much lower than rating for overhead and reliability.

They will run BTL into a single 4 ohm subwoofer at +/-30V rails (25 under load) so each amp will be capable of less than 2 ohms to easily drive the bridged 4 ohm load.

The issue is I don't want to put the drivers on the heatsinks, but want them to stay cool under full load while on the circuit board. I figured a TO-220 may run hot, and have much more Beta-Droop at 2 ohms, so I plan to use TO-3P MJW0281 and 0302 pair for the drivers, and having them drive 3 pairs of MJL4281/4302. I need the heatsink space because 6 total output pairs of MJLs nearly take up the whole heatsink.

I haven't built the amp yet to test, but I'm wondering if a pair of MJW0281 and 0302 can handle driving the low load without heatsinks?
 
gain said:
start by figuring out the maximum power the drivers will dissipate under worst case conditions (maximum anticipated Vce * Ice).

Depending on if I calc w/25v or 30v rails, I get 7.5W or 9W worst case.
:eek:

It looks like they may run hot anyway. Over 5W is a bit much to me. Being they are 150W transistors, you can run them hotter than a smaller transistor. Maybe a piece of aluminum may be needed, I'll just build and see. The OnSemi power transistors as drivers may run cooler than I expect.

*I'll see what I can do first. I really like the idea of just the outputs on a sink, like lots of car amps and HT receivers. Due to the other predrilled holes on the sink, and size of my outputs, I don't have much flat extra surface to use for drivers. I used less outputs on my other sink that's the same as the sub sink, but for the voice amps, to leave room for bridge rectifier and voltage regulator.
 
Hi

They make small individual heat sinks for TO-220's. Always assume the worst case, so is it 9W each driver or for the pair? Would the drivers running hot affect output thermal bias stability? Is there room to parallel the driver transistors? I assume from your brief description you are using a Darlington emitter follower OPS.
 
Actually, I am making CFP output stages. I calculated 9W each driver, and I don't like my electronics running hot. However, RMS calculates at 2.5W, still a bit warm for no sink, but ok.

I do want bias stability, but I'm not setting a really high bias current because it will be a subwoofer amp. I'm more interested in output drive current and voltage swing for this amp. It will be designed with a 40mA or 50mA VAS stage to ensure it can drive the output stage with any load.

After playing with the heatsinks and brackets, I may have found room to mount TO-220 drivers, and possibly even the bigger TO3-P drivers, which I'd prefer. Being there's not much room, that's why I am curious if you can use the TO-3Ps without sinks. I've seen many amps with naked TO-126 for drivers, and TO-220's, but not bigger.

If I manage to squeeze all 12 output devices, and 4 drivers on the heatsink, I barely have enough room for anything else. That makes temp compensation harder, as now I need to figure where to mount 2 transistors for that.
 
Hi

With all those outputs at stake, I would think you want some sort of thermal bias control unless you plan to run with a heavy class B, which may be OK for a sub. Also if you are using CFP output arrangement, it is the drivers than need to be thermally compensated, not the outputs. A large temperature change on the drivers will change output bias significantly. IMHO, CFP is not the best way to drive highly reactive low impedance loads. But that's just MHO.:) Triple Darlington emitter follower might be more appropriate. The downside there is you need a VAS since there is no voltage gain from the OPS and the output transistors would need the thermal sensing.
 
Thanks for the info. I can see why the drivers temp makes a difference in CFP.

****I want to use CFP because it swings closer to the rails than darlington, and a high current VAS to eliminate the need for predrivers, which also helps it swing closer to the rails because of less transistor b-e voltage drops. ****

I designed and built a 350W @ 4 ohm amp CFP w/5 pairs about 2 years ago with great results. It runs +/-75V rails, runs cool, and has lots of raw power, and eats most speakers for lunch. :D:devilr:

This new sub amp will be two amps in BTL @ 4 ohms, and since I have lower supply rails, I want to minimize voltage drop to get the most of the voltage to the speaker, instead of dropping it in heat. I'd also rather load down the transformer 1st, than have my amp dropping voltage or limiting current, or burning up, so I'm building it brute.

The "other" side of this project, the Four fullrange output channels, is a bit easier as far as transistor mounting. I have 4 pairs of *NJL4281/4302 Thermaltrak for my 4 channels on the heatsink. The TO-220 drivers for all 4 channels will be on the PCB. TempCo will be handled by the Thermaltraks themselves. Each channel is only 60 or 70W @ 4 ohms, so no HS is needed for drivers, and I can make the amps compact and simple. I'll bias them real high since they are fullrange, and use the Thermaltraks, to get better SQ. :)

I wish I had more *like* pairs of ThermalTraks, I would have used them for the subwoofer amps as well. Since I don't, I have to come up with a simple solution, since these 2 sub amp channels push more current than the other channels in comparison. I've pondered using the thermaltraks as drivers, but they are big and take a lot of heatsink space, and I'd hate to waste such nice transistors on low power.

*Another brainstorm...........
I do know from experience that a 60W amp won't heat up TO-220 drivers, and being that each sub amp is approx 120W each, maybe having drivers over twice the rating of the TO-220s in each amp would have the same effect..........hmm
 
Hi,
VAS to Driver to CFP outputs feeding the 2ohm load impedance is asking a lot from the VAS.

Peak transient current from 25Vdc (loaded) supply rails into a 2ohm reactive load could be around 35Apk.

What current gain do you expect in the output stage when 35Apk is flowing to the load.

What current gain do you expect in the driver stage when the necessary base current to the 35Apk outputs demands a supply?

What base current will the driver demand of the VAS.

What phase angle will the VAS see when worst case load is presented.
Can the Driver and or the VAS survive the SOAR limits for mS or so that the transient exists?
 
AndrewT said:
Hi,
VAS to Driver to CFP outputs feeding the 2ohm load impedance is asking a lot from the VAS.

Peak transient current from 25Vdc (loaded) supply rails into a 2ohm reactive load could be around 35Apk.

What current gain do you expect in the output stage when 35Apk is flowing to the load.

What current gain do you expect in the driver stage when the necessary base current to the 35Apk outputs demands a supply?

What base current will the driver demand of the VAS.

What phase angle will the VAS see when worst case load is presented.
Can the Driver and or the VAS survive the SOAR limits for mS or so that the transient exists?

Strange, I calculate less peak Amps. I didn't get into phase angles, and all, but I have a feeling it should be solid enough to drive the low impedance loads.

*I'm now using NJL0281/0302 for drivers and MJL4281/4302 for outputs, so my drivers won't be breaking a sweat. I'll be using big MJW0281/0302 transistors for VAS and CCS too:D

50V of rail (BTL) into 4 ohms peak is 12.5A. 60V of rail makes 15A.
It calculates the same with 25V and 30V and 2 ohms each side.

But for overkill sake, lets double that and say 30A.
Outputs:
3 pairs is 10A each transistor. Each transistor has an avg gain of 40 with 8A of current, so close enough.
30A/40 = 0.75A to drive outputs. To make it easy, I'll round up to 1A to drive the outputs.

For Drivers:
It says min of 80 gain for 1A of current.

1A/80 = 12.5mA needed from VAS to run the drivers. Being I'll use a 40 or 50mA VAS, it should drive them fine. If during building, I feel I want even more VAS current, I'll make changes.
 
Update on output/driver configuration................

First of all, thanks for the replies! It helped me rethink and replan the setup, and now it's even better.

I re-cut some new Aluminum clamps, and moved the transistors around so they all fit on the heatsink now. I decided to use ThermalTrak transistors for Drivers anyway, I just used the lower power 180W ones so I had all my 200W ones for other projects. This saves me from mounting temp co transistors and worrying about driver and output temp. I plan to use oversized VAS stage transistors so they don't get hot at 50mA or so, and have plenty of drive.


Transistors for two BTL sub amps:
12 Outputs: 6 MJL4281 6 MJL4302
4 Drivers 2 NJL0281 2 NJL0302
2 VAS MJW0302
2 CCS MJW0281
 
PICS!

Here's the Subwoofer Amplifier Sink and outputs.
 

Attachments

  • subampsink.jpg
    subampsink.jpg
    31.2 KB · Views: 170
Think of a resistor in series with a capacitor - essentially what the speaker looks like above resonance. Now let the cap charge to positive rail. The amp then swings full negative. This gives a peak current that is V(clip)*2/R(e). It might even be worse due to resonance buildup. Then again, this is a very strange signal too. It's probably not very common for this to happen.
 
megajocke said:
Think of a resistor in series with a capacitor - essentially what the speaker looks like above resonance. Now let the cap charge to positive rail. The amp then swings full negative. This gives a peak current that is V(clip)*2/R(e). It might even be worse due to resonance buildup. Then again, this is a very strange signal too. It's probably not very common for this to happen.

Hoping that these are some of the best BJT around, I surely would hope they can take extreme loads. I'll clamp them with diodes like in my other amp so any high voltage on the output drains into the rails.

Have any of you here tryed the Thermaltraks as drivers for regular BJT before?

I've looked around, and I haven't seen any here that do that, so this may be a first. It just makes it so much easier to hook up than a seperate transistor vbe multiplier.

peace
 
megajocke said:
Think of a resistor in series with a capacitor - essentially what the speaker looks like above resonance. Now let the cap charge to positive rail. The amp then swings full negative. This gives a peak current that is V(clip)*2/R(e). It might even be worse due to resonance buildup. Then again, this is a very strange signal too. It's probably not very common for this to happen.

Driving either a pure capacitor or pure inductor puts full rail at peak current on the transistor. But loudspeakers aren't that bad a load - at the impedance minimum, it's always pure real. When the impedance goes reactive, it's ALWAYS above the minimum. Resonance buildup is *high* impedance so you don't have to worry much about that.
Design for an ohm and a half resistive and maybe 3 ohms full reactive within whatever comfort zone on SOA you're going to follow (DC, 100ms, or whatever).
 
I were not talking about pure capacitors or inductors. You are right about the dissipation.

It's just the peak currents that might be that bad, what I'm talking about has no significant effect on transistor dissipation. It's resonant buildup and then a transient in exactly the right (wrong?) place that will demand huge peak currents. It might not be too important, loudspeaker distortion will probably mask any clipping due to current starvation on the transient, and it will also be a very rare occurence.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.