Vbias and Semelab Double Die LMOSFET

[synonymous]

Hello,

Finished up an amp module of Randy Slones' Optimos design.

From what I am told, calculating Vbias is as attaining a quiescent current of 40mA.
I am using the Double Die Semelab devices and .5ohm RE or RMOS resistors. The calculation of Vq is as (40m) (.5+.5), or 40mV. Is this correct??

I ask because I am getting some distortion during medium volume playback. I would guess maybe at about the 100watt range, I get distortion when a sound like a heavy bass guitar or tom tom sound is played back. It is an awkward distortion to me because it appears to be coming out well before the module reaches it's power clip, and, the distortion seems to favor the midbass frequencies. The amp and power supply are capable of 220w at 8 and 300+w at 4 ohm. I am using ON and Fairchild semiconductors, 1% resistors etc...

I suppose the distortion could be a result of some other anomaly aside of Vbias adjustment, however, this is a good spot to start the questions.

Suggestions??

THX
Scott

[Workhorse]

I think you need some nice inputs from our friend Anthony Holton

www.aussieamplifiers.com

I hope he will suggest you soon with something good

[synonymous]

Perhaps I should have stated that I design my own boards for my case based on Randy's designs.

Were my calculations on Vbias anywhere near?

THX
Scott

[Charles Hansen]

Originally posted by synonymous
Finished up an amp module of Randy Slones' Optimos design.

From what I am told, calculating Vbias is as attaining a quiescent current of 40mA.
I am using the Double Die Semelab devices and .5ohm RE or RMOS resistors. The calculation of Vq is as (40m) (.5+.5), or 40mV. Is this correct??

It seems like at least a couple of things are funny. First of all, I know nothing about Sloan's design, but 40 mA is very low bias for these parts. The single-die LMOSFETs from Semelab have a zero tempco point of 120 mA, so the double-die parts should ideally be run at 240 mA. Below that point they will exhibit a positive tempco and the bias will not be stable.

I don't know where Sloan came up with 40 mA of bias. That is just silly. Increasing the bias past the zero tempco point will improve the sound quality, but you have to make sure that the heatsinking is up to the task of cooling the devices. This will largely depend on the rail voltages you are using.

I'm also not sure why you are measuring across *two* source resistors (0.5 + 0.5). The normal procedure is to have a single source resistor for each output device and measure across that source resistor. So it seems to me that you are really only running 20 mA of bias. This is 12 times lower than what I would recommend and could definitely cause distortion.

However, I would assume that the circuit also uses an overall feedback loop. This would tend to lessen the measured distortion, but not necessarily so much the audible distortion. Can you post a schematic?

[synonymous]

Hey thanks,

I at one time had it measured out at up to 120 some mV and seemed to have better results audibly, but didn't trust my guesswork.

I am measuring across points equivalent to "TP3" and "TP4" on the schematic. I also am using the Semelab double die devices instead of the Hitachi as shown.

Of note, the following is verbatim from the project directions..

*R40, R41, R42, R43 / 5-watt, 5%, power resistor & Vbias settings
double die mosfet = .1 ohm @ 18mv VBias / 400 wpc
single die mosfet = .22 ohm @ 30mv VBias / 200 wpc
single die mosfet = ,5 ohm @ 50mv VBias / 200 wpc

However, in his book of projects he described it as 40mA target across the resistive value.

Big thanks for the reply
Scott

[KSTR]

I run Semelab/Magnatec BUZ901D/906D in a bass amplifier and can back Charles' statement that they need at least some 300mA (IIRC) bias each to sound good and keep the bias stable / slightly decreasing with increased temp.

National AppNote AN-1645 even states 180mA (found empirically) for the single die BUZ901/906 and contains useful graphs of bias vs. time after a load of 40W has been switched of (allowing to reach thermal equilibrium before switch-off, of course) which clearly show that the lowish 40mA will cause severe bias drift without further thermal compensation.

- Klaus

[KSTR]

Scott, you write about "40mA target across the resistive value", sure about that exactly, I mean shouldn't it read 40mV instead of 40mA. And further, across which resistor value exactly?

From the values you gave in the cited table I read 100mA and 136mA resp. for single die and 180mA for double die.

- Klaus

[synonymous]

Thanks,

After wrestling a bit and learning of image and file size constraints I have uploaded the schematic image. Hopefully it is legible.

I wonder why such a large difference.
Hopefully this discussion will resolve the issue. I have made other modules in the past and had similar issues and always felt it were me or the parts. This could really crack the code.

THX
Scott

[synonymous]

40 mA was the stated desired quiescent current by way of calculation, I*R, (40mA)*(.5ohm+.5ohm) or (40*1) to of course equal 40mV

[KSTR]

Ahem, you replaced the single die J162's/K1058's with a double die type for each device, leaving all else unchanged?