Aleph X - rootcause of absolute DC offset drift
I have just successfully started up and adjusted my first monoblock of Aleph X (17,5V rails, 6,5A, 8 FETs). Everything ok so far – but I have a drift problem with the absolute DC offset. Maybe some freaks could give me further tricks to reduce it. I searched for this issue in the huge treads – but how to find it... So I start this tread!
The situation – as I understand it: my LM329 voltage reference brings the diff FET pair almost immediately to the final adjusted current, 2 x 11mA. So the biasing voltage at both 392R resistors is stabile after already some minutes. This stabile voltage brings the lower power CCS (Q2/Q11) very fast to a stabile current. It starts with 1,67A and reaches 1,68A per FET already after 10 min. Perfect so far...
But now the problem starts – the upper CCS – why ever – starts with 1,75A! Just after fully warmed up (taking 45-60min) it comes down to the same value of 1,68A per FET to get zero absolute DC offset. 1,75A – 1,67A = 80mA x 2 Fets = 160mA to be sink into R1/R4 with 2 x 100R. This forces 8V of absolute DC-offset directly after switching in! After 45 min still 1V.
Then I installed R48 and R49 with 4kOhm which were laying around. But this just reduces the value of the offset down to ca. 40%. So starting with 3,5V and drifting over 60min to almost zero strongly depending of the final temperature. Lower R values should improve the behaviour.
Lets analyse the rootcause to identify solutions. If my understanding is right the Basis-Emitter-voltage of upper npn Q3/Q8 must be drifting to cause this behaviour. But what makes this thing drifting?
- I cooled the npn, almost no change in offset.
- is it maybe a drift of the V1/V3 poti?
remark: my relative DC offset at the speaker is below 10mV any time, both amp branches work perfect symmetrically – drifting...
How are your experiences about the typical drift? Are there any ideas to compensate this drift if it is normal behaviour? Do R48 and R49 have an negative impact on the sound?
Thanks for your freaky answers, I am a bit of lost!
the variation in absolute dc offset is caused by variation of the dc bias of the diff pair. The tempco of the zener and the fet are not the same and bias current is changing until the temperature is constant.
I tried different zeners (a negative tempco is nice) and coupling of the fet to the case. This allowed for raising the feedback resistors (R46/47) to over 10k for a better sound.
Somewhere on the forum there´s a thread describing the details but I can´t find it at the moment (but I will look)
here it is....
My experience is more in line with Klaus than with William. I use a different (JFET) current source and it stabilises quickly, as confirmed by the voltage across the diff pair drain resistors. The BJT of the Aleph current source is the cause of most drift.
I say this because if I have extremely good thermal coupling of output MOSFETS to each other (unmeasurable temp diff) and to the heat sink (case to sink 2 degC), so the drift of the output FETs largely cancel each other if Vgs stays constant. This I know as I tested each sets of FETs on heatsink at working voltage and current, using 9V batteries and trimpots to bias, to check thermal behaviour. And the current of upper and low FET sets, though both drifting with temperature, remains essentially identical (to about 10mA).
And I can measure much more drift on Vgs for the upper FETs (biased by the BJT) than the lower ones.
PS I do not use any bootstrap caps in the Aleph Current Source.
Both are sources of drift, and also the varying Vgs of the
output devices. Since the simple version of the circuit has
no feedback for common mode effects, the absolute (as
opposed to differential) DC will drift. We use some resistive
loading to ground along with some comon mode feedback to
the Sources of the input devices to reduce this, and then
we adjust the offset after the amp has warmed up. You can
use a "servo" circuit if you like, but we have found these two
techniques plus some patience is adequate.
thanks to all, especially wuffwaff and the one and only, for your hints.
Well, the reasons for the drift are well understood now, the causes and its relations are very complex. I will think about passive corrective measurements, like NTC etc. ... Maybe an improvement can be found. Maybe even an active regulator?
thanks for the link. What are the Mcmillan resistors mentioned in the post? Are these in Hifizens pcbs (Greys original schematic) or should they be added?
the McMillan resistors are the 4k7 between gate and source of the diff pair (R46/47) in Hifizens schematic.
I will try a dc-servo as soon as I find some time....
I believe it's from each side of the Output (+ & -) to the Sources of the diff pair.
An active offset regulator will work fine. If I ever get tired of
adjusting the pot in production, I'll put one in.
Keep in mind that the value of resistors from output to the
Sources of the input diff pair are very dependent on the specific
amplifier - 4.7K is a typical figure, but the overall performance
varies with the values, so you have to test them not only for
efficacy in controlling offset, but also distortion vs amplitude and
frequency and CMRR.
|All times are GMT. The time now is 01:20 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio