Zen instability?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all, comments invited please on the following and how it could be prevented next time. I found a solution, but without understanding why. This seems to be a facet of my implementation rather than a design problem – I’ve not come across it before.

While scoping my Zen Revisited for symmetrical clipping I experienced bad HF instability embedded on the top 20 or so degrees of positive (output) peak of the test sine wave – around 30% of the peak signal in amplitude – i.e. pretty horrible – it occurred only above a certain output level and came on quite suddenly.

I use TO3 MOSFETS from Exicon, plastic washers – don’t know the capacitance to case (drain). The electronics are point-to-point onto the MOSFETs/other big parts, all in fairly close proximity to the sink body. There are two such gain stages in parallel each on it’s own sink, which are adjacent, but not touching each other.

Each sink is grounded to PSU 0V rail via a cable run of maybe 15cm.

= I then discovered that touching either heatsink with my hand killed the instability completely =

On a hunch I snipped the heatsink ground connection to the PSU 0V – sinks still inter-connected but now floating – and all went quiet and well behaved. Phew! So what’s going on here? A clue maybe that I’d just replaced an electrolytic - (but at least it was stable!) - version of the input coupling cap with a foil job – all the wiring around the input was necessarily longer because of the horrendous size of these critters. $64000 question: does it sound better with these caps? Hmm, jury’s still out on that one….

Cheers

Pete
 
I read with interest your comments about the ZEN as I too have been playing around with the design. I cannot help you on your heatsink grounding problem as I am yet to ground mine, I don't seem to have a problem with this(everything is on a big sheet of MDF and not in a case as it should be). You say you replaced the electrolytic input coupling capacitors with higher quality foil type, I was/am considering replacing mine with better quality film type caps but have not been able to justify the expense. Could you let me know what your findings are with this substitution, it may save me wasting money.

Cheers

Dan
 
Thanks guys. I have to build another of these for the other channel - so can experiment.

Jam, yes, I'll try and improve the local HF PSU decoupling which I think is your point. "B+" - not familiar with this notation - you mean the + supply rail??

Ding, yes, it's a bit of a gamble re the 'better' caps - I'm putting my instability down to the additional wiring length, but maybe also the fact that these caps are glued to the back of the sink - so there must be a little capacitive coupling to the sink. The MOSFET drains i.e. the amp output are capacitively coupled to the sink via the TO3 washers. Is this significant? The previous diminutive electrolytics were more or less flying free. Comments... The back of the sink is effectively the chassis of the amp.

Also in the Revisited Zen these caps are in the f/b loop - said by NP to compensate (dependent on amount of f/b) for their own shortcomings. So dunno really about improvements, the amp certainly sounds nice, but I was quite impressed with it before too. Yet to hear it in stereo though - estimate 3 weeks. I may try a few more scientific experiments on the second channel.

Cheers

Pete

[Edited by peted on 07-27-2001 at 04:30 AM]
 
Pete,

Sorry about the confusion. Yes, I mean the + supply rail.

Another thought is to increase the value of the gate resistors of the mosfets.(The gate resistors must be attached as close as possible to the mosfets).

Cheers,

Jam

[Edited by jam on 07-27-2001 at 07:31 AM]
 
Nelson, Jam, many thanks. Bear with me guys - I want to get to the bottom of this...help other's avoid it if possible...

1) There is no gate R shown in the original Zen Revisited schematics for Q1 - what would you suggest - 220ohm as per the current source? Is it simply the inherent lo-pass filter formed by this R with the gate capacitance that is controlling the parasitics?

2) Can either of you explain the oscillation mechanism here and particularly why it is so sensiive to heatink floating (stable) Vs h/s grounded (unstable), and touching/not touching the h/s (when grounded)? My motivation for picking away at understanding this is to be able to avoid the problem when constructing other channel (and fix the first).

3) Thinking about that grounding, it is not direct - 15-20cm of wire. Hindsight instinctively suggests better to strap h/s to the stage's 0V rail via shortest route - maybe only 5mm in my case.

4) Running with floating h/s doesn't feel at all correct.

Sorry - I'm taking the less pragmatic approach because the mechanics of this implementation don't make it easy to play around - I'd never anticipated having to do it!

Cheers

Pete
 
Pete,
Yeah...try 221 ohms. There's lattitude to that value, so if you're out of 221, try 475 or something. Last fall, I was playing interesting games with a baby version of an Aleph output stage and was dropping in values anywhere from 100 ohms to 10k ohms; however, values on the lower end of that range are advised.

Grey
 
Update: Heatsink grounding and instability - lessons learned...

Here's what I did - its behaviour is now impecable

1) Grounded the heatsink direct to (well - 15mm away) the star earth point (my electronics are sitting slap in the middle of the heatsink). Dispensed with previous h/s grounding scheme.

2) Added (across the supply) a 2.2uF polyester and a 47uF electrolytic hard up against the gain stage electronics as local HF PSU decoupling.

3) Added a 220ohm gate resistor to the gain MOSFET soldered direct to the MOSFET gate terminal.

Thanks for the help

Pete
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.