supergreen: I did it!

[..]

Hi Olaf,
to gain switch - negative
to ckt - positive

(i think it's like green mk I, correct me if i'm wrong)

Did you use cheap switches? i use a NOS ITT switch that doesn't give *any* noise at switching...
Try using RMAA (rightmark audio analyzer, free version available) for measuring the noise level at a specific gain (and do some other tests like frequency response etc) so you can compare with the specs of other gear. Needs a good soundcard/AD, anyway...

btw my green mk I /Mozart is *very* clean and sounds very good to my ears (i never expected *this* good quality with my first diy micpre).
However, i get a roughly one times per second hum/noise audible with >50dB that goes away if i shield some hookup wire with my hand, so i will rewire again using shielded wire to make it a really high quality unit besides that i have to check for proper grounding ones more...and maybe change to another/better power transformer and use this one for my first GSSL, where it fits specs and even the pcb )

Olaf, did you keep the 1k resistors at the input? If you don't have this funny HF transformer, you gotta take out the resistors or replace them with very small values. Without the HF transformer coils bypassing them at audio frequencies those 1k resistors will add a lot of noise.

The best mics to test noise performance are dynamics. Ribbons in particular. On many condensers, the self noise is quite high and masks the preamp noise.

The 9098 uses parallel nose-to-tail caps, so polarity is actually both ways.

I have built the 9098 about 18 months ago, but couldn't get it to sound right. (I regularly service four 9098i consoles, a total of 256 mic pres, and 576 inputs!!! :grin: )

Eventually it got back-burnered, because despite giving it a really good effort, it just didn't sound right. Check the lengthy thread that covered what was done, I don't really want to re-type it all right now. :wink:

Anyhow, I visited the Amek factory back in 2002 and spoke at length to Andy Price, who did the 9098i layouts. He said it was tough work. I've also met with and chatted to Rupert Neve about the same stuff... Also, a good number of people also seem to feel that much of the 9098i rack units don't even sound quite right either.

Good luck with it, but after comparing it to the original, I couldn't offer it as a product; it just didn't work. now, the S*S*L 9k is a different beast. After working on that for a short while, that one worked beautifully! I made that one available to all the DIY-ers because it sounds like the real thing and works beautifully well. I think that a good number of people have built them and as far as I know, all seem to be delighted with the result.

My guess is that you're probably bumping into similar 9098i issues to those which I experienced... and I have the benefit of several spare 9098i console modules on my bench to play with and compare things to.

The large caps are crucial, and display nonlinearities at low levels. Too low a value causes LF rolloff at higher gain settings: measure the frequency response at maximum gain and you can see what I mean...

It's a tough nut to crack. It's on my to-do list before I die, but not that high a priority now that I know it's not going to be as inexpensive a product as it appears to be from first glance at the list of components.

Keith

Absolutely delighted !
Thank you Keith to give us the chance to build the 9K ! :green:

Neeeeeeeeeeeeeno

The 9098 uses parallel nose-to-tail caps, so polarity is actually both ways.

Click to expand...

Anybody tried an anti-series arrangement here ?
(say two 10.000uF caps with the center-point to a rail via a
high-valued resistor - or to keep the value make that two 2*10.000uF :? ).
Required cap values get messy of course which will probably be
the main reason it's not used...
But when the highest gains are rarely used I guess all those other
settings could benefit.

Or how about skipping the highest gain-setting (relaxing the required
cap value) & adding those few lost dBs after the first stage ?
Messy as well perhaps, but couldn't imagine noise-penalty would be
that severe right away (for most uses).

Bye,

Peter

Olaf, I just took a look at your supergreen pages (very neat!). I saw that you have an additional impedance switch. I think that if you want to use the lower impedance setting you should use bigger input caps. Only 10u will produce some low cut and associated LF phase shift at the low impedance setting. Just a suggestion. Anyway, I'm curious how your Supergreen turns out.

[quote author="clintrubber"]

The 9098 uses parallel nose-to-tail caps, so polarity is actually both ways.

Click to expand...

Anybody tried an anti-series arrangement here ?
(say two 10.000uF caps with the center-point to a rail via a
high-valued resistor - or to keep the value make that two 2*10.000uF :? ).
[/quote]

Peter, I actually thought the same thing. Those caps at the the gain setting resistor are really important, and notoriously undersized in many budget designs.

I've never seen an acutal 9098 PCB, but I suppose there wouldn't be room enough for 4 * 10.000uF caps. But might be an option for a re-design. On the other hand the physical size might be a problem because of junk that might creep into the audio.

THey're H-U-G-E! -by far and away the largest caps in the I/O module.

4 would mean taking out a mortgage to build something to house the preamp in! :wink:

Keith

Peter posted

"Anybody tried an anti-series arrangement here ? "

Yes

four 4,700uf might work better

http://www.allelectronics.com/cgi-bin/category.cgi?category=140500&item=MKP-204&type=store

20.0uf polypro caps. pretty large though.

[quote author="Gus"]four 4,700uf might work better[/quote]
Actually I reckon that to equal 2 x 4,700µF in counter-parallel (don't forget that's essantially 9,400µF) you'd need two 20,000µF caps in anti-series, with or without a pull-up/down biasing arrangement.

Can anyone fault my math?

Keith

Can anyone fault my math?

Click to expand...

No it's correct; and as also mentioned in my previous post.
Uhh, so that's 0.004F in total to make an 'OK-electrolytic' of 10.000uF ! :wink: :?)

Anyone any thoughts on that mentioned approach to reduce gain
a bit in first stage & compensate in the next ? The cap-requirements would improve so quickly ! :grin:

Bye,

Peter

Hmmm, given the complexity of switches often seen in various gear, why not try that route here as well ?
I mean, couldn't the cost (& height...) of the total unit be reduced by using the anti-series arrangement with more decent (lower) valued caps for all but the highest gains and if one really must have that max gain then the present Green-arrangement (single cap or two caps anti-parallel) is used.
It'll mean an added switch-deck, but the uF-requirement would be greatly reduced.

One could argue however that the highest gain (= the lowest signals which might need the improvement the most) still doesn't get the benefit.

But one could also argue that because of that d*mn highest gain, all other gains have to suffer from that monstercap with all its possible side-effects....

What I meant was two 4,700 in || for about 9,400 uf and another two in || that mess in series. Wired for forming voltage across the caps

All that for a nice 4700uf NP cap

Low ESR caps measured and matched

good luck finding a good 10,000uf cap

OK so we need





:grin:

but yes...

good luck finding a good 10,000uf cap

Click to expand...

:? Those blue ones probably better suited for supplies... :?



But serious, now I'm using a single Elna elco 4700 uF/16V,
not even that big in size. Might add a few more. FWIW, since it looked from some time ago I neatly reformed the thing, only a few uA of leakage left. Dunno about ESR for this one though.

Elco's have large spread, so your LF-filtering could vary quite bit - so could be different between channels as well.

oh damn iread that wrong, i thought that was 10period000uf not 10comma000uf!!

what's the absolute max voltage we need?

Digi-Key Part Number P6975-ND

5.5v 1.0F (yes that is one farad) Electric double layer (gold)

4.82$

have no clue how they sound but i have some .1F (10000uf) versions of these caps i use for memory apps.. i could put one in something to see how it sounds..



Digi-Key Part Number 283-2514-ND

5v 1.0F Aerogel(yes carbon foam!)

9.00$

I am seriously interested in how these sound.. they are supposed to be damn near ideal caps.

10period000uf not 10comma000uf!!

Click to expand...

I'm wondering where exactly in the Atlantic ocean it's being switched over :roll:

I mean, those periods & comma's and also that confusing date-sequence - like: is it the 6th of July, or the 7th of June ?! :evil: :? :wink:

Bye,

Peter