Im am currently doing some tests on a commercially available mosfet (car) audio amplifier, and I'm having some doubts as to the correct bias setting procedure . It's a classical Class (A)B amplifier with FQP22P10 and FQP33N10 mosfets and 0.22 Ohm source resistors.
I am using a HP8903B distortion analyser, measuring distortion at a 4 ohm load. But what is a good compromise between THD and power consumption, besides the effect of thermal runaway? And what would be a good output level for this measurement? 1 Watt? And what frequency, 1 Khz?
Without any load attached I can make crossover distortion dissapear with modest biasing currents, but with a 4 ohm load this is not possible so I have to find some compromise.
I tried searching the web for your average mosfet biasing basics but failed in finding any. Offcourse I am still interested to receive any interesting links..!
I am using a HP8903B distortion analyser, measuring distortion at a 4 ohm load. But what is a good compromise between THD and power consumption, besides the effect of thermal runaway? And what would be a good output level for this measurement? 1 Watt? And what frequency, 1 Khz?
Without any load attached I can make crossover distortion dissapear with modest biasing currents, but with a 4 ohm load this is not possible so I have to find some compromise.
I tried searching the web for your average mosfet biasing basics but failed in finding any. Offcourse I am still interested to receive any interesting links..!
For a right class A working, the quiscient drain current must be equal or greater than the maximum expected load current, i.e VDD/Rload. Did you check this?
1 watt, 1 kHz, 0.01% THD. What bias does that need? Has it just one pair of output devices? Then 50-100mA would be acceptable.
I'm sorry, I must be expressing myself wrong. I don't want to run the amp in class A, I want to apply just enough quiescent current to make the amplifier run in class B (or AB) without excessive crossover distortion.
The reason I posted "class (A)B" is that I thought that this was called class AB, though Douglas Self clearly states in one of his books that an amplifier that has just enough quisecent current to avoid crossover distortion is called "class B".
Anyway, in my opinion, the default quiescent current setting for this amp was off causing high levels of crossover distortion. So the question is, how to go about when setting the quiescent current for this mosfet amp?
I wish 😀
I could run it in to class A and the analyser still wouldn't read 0.01%.
Have to say that I don't use an "A weigthed" filter, still looking for one. I will check tomorrow with the 30 KHz filter enabled.
The amp uses two pairs of output devices for each channel at 48 volt railvoltage (96 rail-to-rail). At 100 mA this would be 4,8 watt for each output device. 4 mosfets per channel, 4 channels, means 76,8 watts of power consumption. Add the SMPS and protection circuitry and the total current draw at 14.4 volt exceeds 6A. I don't think this is possible in terms of thermal stability.
Fresh out of the box the current draw for this amp is around 1,5A.
The reason I posted "class (A)B" is that I thought that this was called class AB, though Douglas Self clearly states in one of his books that an amplifier that has just enough quisecent current to avoid crossover distortion is called "class B".
Anyway, in my opinion, the default quiescent current setting for this amp was off causing high levels of crossover distortion. So the question is, how to go about when setting the quiescent current for this mosfet amp?
I wish 😀
I could run it in to class A and the analyser still wouldn't read 0.01%.
Have to say that I don't use an "A weigthed" filter, still looking for one. I will check tomorrow with the 30 KHz filter enabled.
The amp uses two pairs of output devices for each channel at 48 volt railvoltage (96 rail-to-rail). At 100 mA this would be 4,8 watt for each output device. 4 mosfets per channel, 4 channels, means 76,8 watts of power consumption. Add the SMPS and protection circuitry and the total current draw at 14.4 volt exceeds 6A. I don't think this is possible in terms of thermal stability.
Fresh out of the box the current draw for this amp is around 1,5A.
Last edited:
I believe most if not all car amplifiers run very low or even zero bias as they are made for output power rather than sound quality as most car audio is just *BOOM**BOOM**BOOM**BOOM* anyways.
Just in summary, class AB is indeed the conventional title for low-bias amps (essentially Self's class B).
The figure of 100mA applies to each complementary pair of output devices, since they are effectively in series. 2 pairs require just 200mA.
Anecdotally, a figure of only 70mA /pr may be adequate depending on just how good a sound quality you require.
Are you just referring to idle/quiescent current when you say 1.5A "out of the box"?
Last edited:
As far as i understood it be meant what the amp draws from the 12V input.
That 1.5A would be pretty much what the DC/DC converter alone needs to even operate at all.
That 1.5A would be pretty much what the DC/DC converter alone needs to even operate at all.
I guess you aren't very far from the truth, current consumption, thermal stability and physical size will always make these amplifiers to a compromise. However, there are some very well designed amplifiers that sound well and have good specs.
Last edited:
That is indeed what i meant, except that we measure at 14.4V which is the highest possible alternator voltage.
With the quiescent current set to zero on all channels, current draw drops 200 mA. At low power the DC/DC converter has excellent efficiency, so >90% is used for biasing. This would come down to some 3,4 mA of quiescent current for each pair of output devices.
Hi,
As Self points out, there is no optimum bias current for FETs like there is for BJTs.
Consequently fairly high standing currents are generally preferred, but in the
general scheme of things bias for a FET output stage is really anything you want.
rgds, sreten.
As Self points out, there is no optimum bias current for FETs like there is for BJTs.
Consequently fairly high standing currents are generally preferred, but in the
general scheme of things bias for a FET output stage is really anything you want.
rgds, sreten.
If it can't do 0.01% then it's a pretty poor amp. Please name and shame 🙂
Seems as though it's probably best just leaving it as the manufacturer set it.
Seems as though it's probably best just leaving it as the manufacturer set it.
This is the info I was looking for, thank you!
I've done some measurements and I can get the amp to do 0.033% of THD @ 1 Khz, 1 Watt, 4 Ohm.
Although this would be ok for me the increase in current consumption is 1A for each channel, so 4 amps total. This comes down to around 160 mA of bias current, or 80 mA for each pair of output devices. The biggest problem is probably that the output devices are not matched and bias is not evenly distributed between the two pairs.
Furthermore, increasing the total power consumption by 4A would make the amp go thermal runaway, since the bias network is not compensated for temperature increase.
I would but I can't, since I am a friend of the supplier (not manufacturer) who asked me tho check the amp for flaws.
I disagree, and you would too, if you knew the default setting 🙂
This is what I have been waiting for. Your real intention.
As there is no standard, why not compare with other amp of the same price range? Reliability is also a trade-off for quality.
If you're looking for design flaws, I think you should not look at THD. But of course you have to check if written specs are true. But often specs are not written completely so it is usually not a useful reference either.
- Status
- Not open for further replies.