How hot would depend on supply voltage and current.
Say you use 2x18 volt.
The op-amp itself (no load) takes 5 mA x 36 V = 0.180 Watt
Resistor bias the output to 10mA = 10mA x 18 V =
0.180 Watt / single OP-amp package
0.360 Watt/ dual OP-amp
Will give a total power to dissipate of 0.360 / 0.540 Watt
This is my opinion lot for a typical 8 pin plastic DIP IC.
Running same at 2x12 Volt gives 67% of this power consumption.
Now OP-amps perform the best near their max allowed voltage.
So if possible we should run at 2x18 Volt or more.
Small signal transistors are used in output stage of OP-amps.
Typically for most discrete small signal transistors is
they do not perform their best at currents above like 5 mA.
Around 0.5-3 mA is a usual workspan of such devices.
I would guess the same goes for output stage transistors in OP-amps.
Say you use 2x18 volt.
The op-amp itself (no load) takes 5 mA x 36 V = 0.180 Watt
Resistor bias the output to 10mA = 10mA x 18 V =
0.180 Watt / single OP-amp package
0.360 Watt/ dual OP-amp
Will give a total power to dissipate of 0.360 / 0.540 Watt
This is my opinion lot for a typical 8 pin plastic DIP IC.
Running same at 2x12 Volt gives 67% of this power consumption.
Now OP-amps perform the best near their max allowed voltage.
So if possible we should run at 2x18 Volt or more.
Small signal transistors are used in output stage of OP-amps.
Typically for most discrete small signal transistors is
they do not perform their best at currents above like 5 mA.
Around 0.5-3 mA is a usual workspan of such devices.
I would guess the same goes for output stage transistors in OP-amps.
Hi lineup,
Whoa, let's just look again at this dissipation. If the chip output is being biassed into Class A with a pull-down R ( or pull-up or CCs) surely the other half is turned off as 5mA is well in excess of output stage quiescent. So you only have the active half dissipating. If it represented 2mA of the total chip Iq then there's 36mW now off and the active half was already dissipating 2mA so is now going only 3mA harder, another 54mW so effectively dissipation goes to -
198mW /396mA for 2mA std quiescent.
Cheers,
greg
Whoa, let's just look again at this dissipation. If the chip output is being biassed into Class A with a pull-down R ( or pull-up or CCs) surely the other half is turned off as 5mA is well in excess of output stage quiescent. So you only have the active half dissipating. If it represented 2mA of the total chip Iq then there's 36mW now off and the active half was already dissipating 2mA so is now going only 3mA harder, another 54mW so effectively dissipation goes to -
198mW /396mA for 2mA std quiescent.
Cheers,
greg
amplifierguru said:
...we probably wouldn't be having this conversation
my guess is that many op amps use < 1mA output bias - open loop output Z of 50-100 Ohms is common: 1/2*(26mV/Iq) => 130 - 260 uA
opa134 claims 10 Ohm open loop output Z; => 1.3 mA, presumably output stage bias could be tested by plotting supply current vs output current
I can't see the point of "class A" biasing the input op amp in a multiloop/buffered arrangement - it only takes a uA or two to drive the output amp, surely 100:1 Iq:Iout is "class A" enough
It does seem likely that the newer dsl driver chips run the output considerably hotter - you don’t get -80 dB distortion at 1 MHz into 25 Ohms with lots of crossover distortion for your limited feedback to fix
LT1795 cfa with settable bias current:
Attachments
Hello,
Another point I was thinking of, and somebody mentioned this in another post awhile back, is that OPA627 has superior LF PSRR (<200HZ) to a AD8610. If I recall correctly, the AD8610 has an OLG to 10KHZ and better transient response. It takes time to "rev" up at LF.
Cannot remember the exact technical reason I am afraid!
Another point I was thinking of, and somebody mentioned this in another post awhile back, is that OPA627 has superior LF PSRR (<200HZ) to a AD8610. If I recall correctly, the AD8610 has an OLG to 10KHZ and better transient response. It takes time to "rev" up at LF.
Cannot remember the exact technical reason I am afraid!
Re: Hotter than hell? Are we joking or what???
0.4 W into a DIL08 is pretty hot according to me and certainly the chip itself will be very hot.
Besides "more" isn't automatically better. This "class A current" has an optimal value which much be determined by measurements and also by knowledge how the output stage is designed. I'm sure Scott Wurcer (user here) can answer this because he has designed the AD797.
8 mA typ @ +- 15 volts = 0.24 W (rather warm) + 10 mA@ 15 volts => 0.39 W = pretty hot! On top of that you want to take out a signal I'll suppose.carlosfm said:Are you sure, P-A?
Because it just gets slightly warm.
Sometimes I don't know what's on your mind, really.
0.4 W into a DIL08 is pretty hot according to me and certainly the chip itself will be very hot.
Besides "more" isn't automatically better. This "class A current" has an optimal value which much be determined by measurements and also by knowledge how the output stage is designed. I'm sure Scott Wurcer (user here) can answer this because he has designed the AD797.
As I get it:amplifierguru said:
that 10 mA x 18 volt over comes from OP-amp, at 18 volt of 36 V
and give a power of 180 mW over OP-amp.
If OP-amp, no load, consumes 5 mA x 36 V = 180 mW
to keep all of its circuits going
I think your 18 mW (198-180) seems too low.
Even if half the OP-amp's circuits are turned cold by bias an output 10 mA to negative supply
still would give us 90 mW to keep running
and this added to 180 mW to keep 10 mA over resistor
and a total of
270 mW for single OP
540 mW for dual OP
-----------
In a OP-amp driving constant current source + buffer within feedback loop arrangement
Walt Jung has made this 1mA current source optional.
( se my first post )
Why he added this option at all, I am not sure.
As told here, almost whatever OP-amp would easily drive/bias the input of a buffer, while using true class A though output transistors.
As somebody pointed out, it is only one half of the outpur stage that conducts the bias current, so we should only calculate with a single rail voltage. 10 mA bias and +/- 18 V rails, thus means we get 10 mA * 18 V = 180 mW. An 8-pin DIP has a typical Tja of 100 K/W, which means the chip temperature will raise 18 K (or 18 deg. C). Not a lot to worry about. Transistors are frequently run much hotter in discrete designs.
Christer you must add the normal power consumption also plus the load including feedback.Christer said:
I remember that when I used a bunch of 8-10 mA opamps it became pretty hot when I had 42 of them
My advice is simply that "more" isn't always better. I may be a better strategy to use an active current source. By that you'll get a larger voltage swiwng in class A and possible also lower distortion of the output stage.
As a general rule it's good to keep all parts as cool as possible.
peranders said:
Christer you must add the normal power consumption also plus the load including feedback.
I remember that when I used a bunch of 8-10 mA opamps it became pretty hot when I had 42 of them
Of course on must also consider all other aspects. I was just pointing out that the biasing itself does not cause any extreme temperature increases. However, if the op amp has heavy load or operates in an evironment that is already hot, then biasing might add too much heat. It also obvious that it will be worse for dual or quad op amps if they are all biased, since my calculation was for a single one.
I agree it is good to run components cool, but there is no need to overdo it either. If you are well below the max rating for semiconductors, there is hardly any problem. If you have a Tj of 30 or 50 deg. C hardly matters if the max. is 125 deg. C.
is it wise to do the bias thingy on every OPA627? I mean, I use them in my DAC for I/U conversion and buffering/filtering.
I tried also some other chips but like Carlos said, when you bypass them well...well, they are sounding superbly. If I exchange them with an AD8610 the whole DAC becomes a bit sharp sounding.
I tried also some other chips but like Carlos said, when you bypass them well...well, they are sounding superbly. If I exchange them with an AD8610 the whole DAC becomes a bit sharp sounding.
...hot or not... silicone is tough....
Chip temperatures up to 125°C usually allow full life time.
Even if we can guess that the chip inside might be another 20°C...30°C hotter than the plastic... As long as you cannot cook water on its top, it should survive the thermal stress.
In power electronic chip temperatures up to 150°C are not so unusual and often the limiting reason is not the silicon, but the sealing properties of the potting epoxy.
If such temperatures offer best performance... that's another question.
BTW:
There might be another good reason not to run your electronics at high temperatures all the time. Our human bodies.
Hot electronic tends to fumigate more or less toxic stuff.
The used materials are not optimized by biological background, but by the lowest price for the required performance.
On the other hand... I already survived 20 years of DIY and others survived over 70 years electronic career...
Chip temperatures up to 125°C usually allow full life time.
Even if we can guess that the chip inside might be another 20°C...30°C hotter than the plastic... As long as you cannot cook water on its top, it should survive the thermal stress.
In power electronic chip temperatures up to 150°C are not so unusual and often the limiting reason is not the silicon, but the sealing properties of the potting epoxy.
If such temperatures offer best performance... that's another question.
BTW:
There might be another good reason not to run your electronics at high temperatures all the time. Our human bodies.
Hot electronic tends to fumigate more or less toxic stuff.
The used materials are not optimized by biological background, but by the lowest price for the required performance.
On the other hand... I already survived 20 years of DIY and others survived over 70 years electronic career...
carlosfm said:
Also worth noting. A resistor load to one of the rails maintains a constant power dissipation in the output transistor of the op amp.
That is, when Vout goes up, the current increases inversly to the voltage accross the output transistor.
With current source loading, this is not the case.
A current source loaded op amp would likely have worse settling numbers or more thermal tails.
Mike
Don't know which one sounds better, but I wouldn't be suprised if a resistor won.
Interesting observations.
Where is to find the simplified schematic of the AD8610 and AD8620 ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.