IIRC there is not much benefit from more than 5% Ic matching of the LTP, after that the 3rd harmonics dominate and don't cancel. In most of the designs I see here the LTP is very lightly loaded to minimize distortion from the VAS+OPS, so I wonder if its own contribution to THD is significant.
there's the classic http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/transistors/AN-222.pdf
somewhere I think I've seen a suggestion for audio to trim the diff pair current balance/degen R? for minimum distortion - then inject external offset adj if needed
somewhere I think I've seen a suggestion for audio to trim the diff pair current balance/degen R? for minimum distortion - then inject external offset adj if needed
How did you get the 2N5551 thermally contacted to the pre-driver? I would use TO126 Vbe multiplier and TO126 predriver mounted together.
I'll have to sdd the 2 VBE sensor variation to the book.
Cheers,
Bob
Hi David,
If the trimmer is done in an optimal way, where only the smallest current necessary is passed through the pot wiper to give the trimmer adequate authority, then I guess the only advantage might be that the pot used on the current mirror emitter resistor circuit has very little signal on its pins - the whole pot is largely at an AC ground. When used in the tail of an LTP, the entire pot has the common mode signal on it, which is essentially the input signal in a conventional amplifier design.
I would avoid having to use such a trimmer in either location.
Cheers,
Bob
I use TO-92 in Predriver (thermal coupling TO-92/TO-92), also do this in the current mirror.
The only ones TO-126 that I would use would 2SA1381/KSA1381, but they are not available to me here. My second option would be to use MJE243/253, but the datasheet reports a difference in gain +- 120% NPN for PNP, in my case, I found best TO-92.
Predrivers and drives or only drives even heatsink in the output transistors, if you use ThermalTrak, diode sensor compensate at the same temperature as the internal transistor, Predrivers/drives or drives, give error in compensation due the different temperature mounted even heatsink output TRs.
Now understand why chose to ride on separate heatsink for Predrivers and drives.
Yes but current will be highly temp dependent compared
to the zener referenced current source.
Hi Rafael,
TO-92 devices are OK for VAS and predriver in smaller amplifiers, but in general it is better to use good TO-126 devices for these locations. The 1381/3503 are great devices, I agree.
Cheers,
Bob
Self's data (5th edition p 81) shows small distortion increase even with 0.5% imbalance and and he comments 2nd well above 3rd even with 2%.
Have you incorrectly recalled 0.5% as 5% perhaps?
Best wishes
David
Hi kgrlee,
The EKV models have not been put up on the website, but only because I didn't get to it. I had intended to put them up, but got distracted and that was awhile back (well more than a year). I'll have to go back and see what I've got. I think I did EKVs for all of those devices you listed.
Cheers,
Bob
Hi David,
I wonder if Self's assertion of that high a sensitivity to LTP imbalance is right. It should be something that is easy to sim...
Cheers,
Bob
Been there, done that, Self is wrong. 2nd harmonic distortions are not a good reason to e.g. match the input devices or to balance the LTP beyond the regular component tolerances.
Completely untrue. A 10V zener, for instance, has a positive temp. co. of about 6mV/deg. celsius while a BJT has a negative temp. co. of base-emitter voltage of only 2mV/deg. celsius. So the BJT has a more stable voltage with temp. than a zener diode, unless you use a temp. compensated zener diode.
Last edited:
For the BJT thermal variation is 100 x 0.002/0.65 = 0.307%/°C
while for the zener it will be 100 x 0.006/10 = 0.06%/°C ,
keeping in mind that the CS current is about linearly dependent
of theses temperature variations the zener referenced CCS
stability can be improved when using a 5.6-6.3V zener.
Bateman, in his "Capacitor Sound" series, claims you can get bi/non polar electros with full oxide thickness grown on both foils - and that they measure better on distortion
"back-to-back" construction would have one pair of foils with only the spontaneous oxide layer which is thin so that it sees large V/m - leading to greater measured distortion
even adding a polarizing V to the midpoint doesn't reduce the thin layer's contribution to distortion to the same level as full thickness oxide grown on all foils
http://www.cde.com/catalogs/AEappGUIDE.pdf suggests forming is an external operation - before slitting, winding
"back-to-back" construction would have one pair of foils with only the spontaneous oxide layer which is thin so that it sees large V/m - leading to greater measured distortion
even adding a polarizing V to the midpoint doesn't reduce the thin layer's contribution to distortion to the same level as full thickness oxide grown on all foils
http://www.cde.com/catalogs/AEappGUIDE.pdf suggests forming is an external operation - before slitting, winding
Last edited:
Hi, as i'm sure you're awarelots of non polar Caps made for Xovers are actually back to back electrolytics in a single case, especially high value uf.
Do you consider such Caps as distortion creators too, or just DC unbiased single electrolytics ?
All capacitors generate some distortion. The amount depends on the voltage across the cap. If this voltage can be reduce by clever design then the distortion contribution can be made insignificant. You do need to know the distortion level of the cap first and that you can measure by putting a large voltage across the cap say +20dB with or with out dc bias. Reduce the voltage in steps and log the distortion. Then you know what you're working with. Yes it matters in speaker crossovers too but the best low distortion caps in the values required would become the dominate cost of the system.
Be that as it may a variation of 0.3%/ deg. C is insignificant especially when you consider that the output impedance of the amplified negative feedback current source is at least an order of magnitude greater than that of a zener diode biased current source.