2sa1015* I wrote 2sa1815.
Should've mentioned they are in circuit ( audio amp) at 2.5mA . Whole amp output is 0.5 - 1mV dc offset. I have to try to measure just the LTP outside the cirucit if it's any difference ??.
Today I just finished my amp and was surprised by such low dc offset.
Orher amps I've built were in the 10mV range, didn't bother to match LTP and current mirror transistors so closely tho.
Should've mentioned they are in circuit ( audio amp) at 2.5mA . Whole amp output is 0.5 - 1mV dc offset. I have to try to measure just the LTP outside the cirucit if it's any difference ??.
Today I just finished my amp and was surprised by such low dc offset.
Orher amps I've built were in the 10mV range, didn't bother to match LTP and current mirror transistors so closely tho.
Hi xXBrunoXx,
A perfectly matched pair will make the DC offset close to the calculated value as long as everything else is working properly. This will depend on tail current, base resistors, beta and circuit configuration. So depending, on the circuit and those factors, a 0 mV DC offset may not indicate a matched pair.
Again, the problem with measuring beta individually is that the transistors are not at exactly the same temperature. This absolutely must be the case. Both vBE and beta vary greatly with temperature. Beta is a more sensitive test, so matching that should give you the same vBE as well.
There isn't an easy, fast way out. You either do it correctly or accept the large random element involved. From experience the random part dominates, which is why I thought about the problem and came up with that circuit. It is simple, effective and bang-on. I'm not selling them, the design was given to everyone.
A perfectly matched pair will make the DC offset close to the calculated value as long as everything else is working properly. This will depend on tail current, base resistors, beta and circuit configuration. So depending, on the circuit and those factors, a 0 mV DC offset may not indicate a matched pair.
Again, the problem with measuring beta individually is that the transistors are not at exactly the same temperature. This absolutely must be the case. Both vBE and beta vary greatly with temperature. Beta is a more sensitive test, so matching that should give you the same vBE as well.
There isn't an easy, fast way out. You either do it correctly or accept the large random element involved. From experience the random part dominates, which is why I thought about the problem and came up with that circuit. It is simple, effective and bang-on. I'm not selling them, the design was given to everyone.
Hi xXBrunoXx,
Correct.
There are circuits that have designed in DC offset over 200 mV! (poorly designed!!!). Many may have 20 ~ 30 mV DC offset with everything perfectly matched. Then there is the question of gain at DC. You could have an offset of 5 mV and if the circuit had 40 dB gain (easy numbers, 40 dB = x100) it would measure 500 mV at the speaker terminals due to the circuit not rolling the gain off at DC. Many amplifiers have gain set anywhere from 28 to 35 dB.
If the circuit is designed properly, base current will be low, impedances not too high and the gain at DC is unity, you should see near 0 mV with a good matched pair with matched base impedances. Otherwise you can calculate with reasonable accuracy what the designed in DC offset is. If your pair is matched very closely, your actual DC offset would be close to this. We want 0 mV DC offset, but what we really want is the pair to be closely balanced for the lowest distortion and greatest DC stability.
Correct.
There are circuits that have designed in DC offset over 200 mV! (poorly designed!!!). Many may have 20 ~ 30 mV DC offset with everything perfectly matched. Then there is the question of gain at DC. You could have an offset of 5 mV and if the circuit had 40 dB gain (easy numbers, 40 dB = x100) it would measure 500 mV at the speaker terminals due to the circuit not rolling the gain off at DC. Many amplifiers have gain set anywhere from 28 to 35 dB.
If the circuit is designed properly, base current will be low, impedances not too high and the gain at DC is unity, you should see near 0 mV with a good matched pair with matched base impedances. Otherwise you can calculate with reasonable accuracy what the designed in DC offset is. If your pair is matched very closely, your actual DC offset would be close to this. We want 0 mV DC offset, but what we really want is the pair to be closely balanced for the lowest distortion and greatest DC stability.
The only time and the only way that the amplifier responds only to the voltage difference between the + and - inputs is when the two Vbe/Ic characteristics and the currents are equal. If not, the whole G/(1-GH) equation for closed loop transfer function becomes less accurate. This results in more distortion than expected for the amount of loop gain achieved. For distortions around .05%, these errors can be ignored. If you want it better than that they cannot.
Vbe will change by -2mV/C and the thermal resistance of a TO-92 is ~250C/W. For low offset, the dissipation of the two transistors needs to be equal and low.
Ed
Some isolation and protection from direct air currents will help keep the two packages at the same temperature. I use thermal compound and heat shrink tubing. The thermal resistance between die and case can't be helped, which is why dual transistors are sold on the same die. These are a very good option. Otherwise try to isolate the two transistors from the environment as reasonably as you can. That's in addition to getting a very close match.