HELP!!! Valveaudio Amp

[mgm2000ro]

I had try to make Valveaudio Amp +/-70V Mosfet Power Amplifier (the project name)
this is the link to the pcb for it : http://valveaudio.tripod.com/images/...s/PWCRESCE.PCB (first save as htm then rename as .pcb)
This is the schematic for it:


In the pcb there are some mistakes:
- the terminals of the prefinal are G-S-D, are they corect in this order or am I wrong? I conected the D to + or - and S to the base of the finals? Is it good in this combination?
What is this schematic missing to make it work

PS: Plese excuse my engliesh

[Bensen]

Hi mgm2000ro,

You should do a search on "Crescendo". I uses pretty much ecxactly the same diff and VAS setup.

The connections of the two MOSFET's seems OK for me. But I'm wondering how you can bias these two FET's. Shouldn't there be a resistor between the two sources.

Greetz
Ben

[mgm2000ro]

Yes,this is the problem.
In the original project there is not such thing between the two sources of mosfet,or mosfet and output.
The prefinals J162-K1058 are insensible to temperature (lateral mosfet).
Help me to modify this schematic to be operational.
The amplifier's VAS setup in crescendo configuration is good,there are not problems,everything is functional till the prefinals.

[Bensen]

Yes I know these are lateral FET's, I have in all my design this far used lateral's. You should calculate the max current throught the output stage, this way you can calculate the total base current needed to drive the 4 output transistors at full power.

When you search on this forum you will probably find some kind of rule which will tell you how much the bias current throught these drivers (K1058/J162) should be. For example: bias current throught drivers is 2x max base current for output transistors at max power.

Can you do a simulation with a resistor between the two sources?
Do you have an idea why they used two FET's instead of two transistors for drivers? Maybe to decrease the load for the VAS??

Greetz
Ben

[djk]

There should be two source resistors to the output node. Due to the lack of bias temperature sensing the drivers and outputs should be run class AB+B (as per Crown amps). Try 3R9 resistors with the bias pot set to give 390mV across each. This will run the drivers at their zero tempco point and drive an 8 ohm speaker up to about 150mW in class A.

[lumanauw]

DJK is right. You must put R from sources of K1058pairs to output. The pin of K1058 is GSD not GDS.

The bias should be changed from 200VR to vbe multiplier.

It is based on Crescendo amp. Unfortunately, long time after I draw this, I found out that the first crescendo is prone to oscilation (some even call it oscilator ). You can try to look for Crescendo Millenium edition, it should be more stable, they have RC between the collectors of the differential input.

I'm surprised by you. Not so many people are willing to re-draw the schematic from the pcb. I never include the schematics for the pcb in my website.

You can look at the schematic in my website. The 200W amp or 400W amp (runs from +/-70V) should be better for now days.

[mgm2000ro]

thankeyou lumanauw for the reply
I an kind lost in the sugestions, is this change good?

You can look at the schematic in my website. The 200W amp or 400W amp (runs from +/-70V) should be better for now days.

Do you suget by this noth to use enimore the (K1058/J162) because in your project 200w/400w the prefinal are bipolars

I have tested with one rezistor conected betwen the 2 sources of the (K1058/J162) with an R of 180 ohm, and noth conected to the output.

If it is posible to make an drawn schematic of the changes that are required , I will be most apreciative.

[darkfenriz]

Your schematic causes an ache in my neck when I try to see it upside up and downside down. I suspect I am not only one.

I think you may need frequency compensation for your amplifier. Like caps between long tail pairs resistors or base-of-common-emitter-to-collector-of-common-base (miller) caps in stage of voltage amplification.

[lumanauw]

Hi, Darkfenriz,

Neck-ache eh..? Me too

mgm2000ro, you got another good suggestion. Read Darkfenriz's reply above, combine it with DJK's it should fix your schematic pretty much. Don't forget to change bias scheme to VBE multiplier.

Here's Crescendo millenium. This is more stable than the first released Crescendo amp. They have change the biasing scheme to VBE multiplier. Notice the RC network on differential's collector.
For another stabilization, like Darkfenriz suggest, cap(pf) between base of Q5 to collector of Q4.

[mgm2000ro]

Hello everybody!

Let me tel you what i did:

I tryed not to use the finals 2sc-2sa and i let as final tranzistor J162-K1058(my prefinals). I have put caps (4.7p) between base and collector (T7-T8).I have put the RC (330 ohms,1n) network on differential's collector (T1-T2).
The bias is not changed to VBE multiplier, it remained VR 250 ohms.
The bias is 100 mA, at +/- 60V.

The problem is:
*Sometimes works, sometimes doesn't;
*When is not working the bias is rising much and my amplifier whisels;

I puted between the sources and the output of the finals 3.9 to test the viability of the schematic.

To MR lumanauw:
*What do you recomend me?
1:To change the bias (VR250) to VBE multiplier with 2SK537;
2:To change in the entire schematic the values of all rezistors from Crescendo-82 to Crescendo millenium.

P.S.:
I did't find 2SK537 or 2SK792 - 2SK1643 for bias.
I intend to use 2SK2750.This is good?