5003 eq building and help.

[desol]

Well, i took a few measurements. Picked a random pot from the pile and it measured in at 9.31k(half of which would be 4.655k). At center indent, from wiper to one side of the pot(clockwise), it measured 4.61k. From wiper to the other side of the pot measured 4.87k.

 

[Winetree]

Any leads on those on-off-on switches?
Did anybody use the "audio Maintaince" pots?

 

[peterc]

Any leads on those on-off-on switches?

That was an error by the poster, the switches are on-on

I used Omegs for the prototypes and my builds, which, as far as I know, are the same ones Colin sells.

 

[alexc]

I just finished the first of a pair of dual channel eqs with PeterC 5003 pcbs.

Thanks kindly to PeterC for making this possible 

I did a 1RU rack job as I don't use 500 series racks. I find them way too small for me

Used gar2520 DOAs from ClassicAPI ( which worked first time, running nice and cool  8)
Used a JLM 3 rail PSU with the phantom rail adapted to be a variable +ve (11V) for powering
leds and relays.

Added relay bypass, as I always do.
Added Haufe input transformers as I like to do. Also kept the 5534 based unbalancer for comparison.

Used Ed Anderson's 2503 style ouput transformer in one pair from ClassicAPI, and Cinemags  in the other.
Used PeterC inductors.

---

Well - the good news is that it all worked fine and showed all the right gains and eq curves

Unity gain noise floor with EQs engaged and flat came in at very very respectable -82dBu for the channel furthest from PSU and -80.5dBu for the channel closest.

The dBu figures are RMS 20Hz-20KHz using a sound card RTA (Motu 2408MkII) that is reasonably
calibrated using true rms meter and calibrated CRO as well as the RTA offset calibration feature.
I think it is accurate to a coulple of dBu. My 'loopback' noise floor is -83.5dBu.

So, all good so far.

 

[alexc]

So doing some further testing ...

Did the noise optimisation - rotated toroid for min noise which made an enormous difference,
like around 18dBu of noise in the 50Hz and several harmonics area.

Checked for sensitivity to wiring placement - almost no sensistivty at all, even to my rats
nest of wires 

Checked for sensitivity to grounding wires - no sensitivity at all. I did not use the pcb onboard 22R chassis-circuit gnd resistor as the JLM psu already uses a 10R//0.1uF network for that purpose.

So, all in all very good noise stability, only the toroid rotation making any difference.
Final unity noise floor around -81dBu more or less.

I checked for the differences in using the unbalancer vs transformer. There was a 2dBu improvement in noise floor with the transformer, and a worsening of the low freq -3dB cutoff
from 20Hz to around 30Hz with the transformer. The 5534 was completely flat, the transformer showed a tail off of around -4dB at 20Hz. No other dfference really.

---

The major issue, as expected, is that the inductors are susceptible to picking up 50Hz and harmonics from the toroid when the mid and low band are set to cut or boost and to a lesser extend, the high band is set to cut or boost.

The proximity to channel B to the psu worsened the basic susceptibility to induced noise in the mains frequency and harmonics by around 9 dBu for each of the bands, with max boost/cut in any of the bands.

Adding 12dB boost or cut reduced the noise floor for the best channel from around -81dBu to around -50dBu for the low band, -63dBu for the mid band and -72dBu for the high band.
Similar for cuts of 12dB for each (about 3dBu or so better for cuts over boosts).

On top of that, the channel closest to psu was around 9dBu worse.

The changes in noise floor happen very rapidly when moving off the centre 0 gain position, which is also the noise floor 'null' in each band.

Simple shielding tests using a couple of folds of sheet iron for around 3mm held nearby and on top of the inductors  indicated maybe 5dBu or so improvement for worst channel, maybe 2dBu for the best.

So there it is so far in my tests - unity, flat eq noise performance is stellar, gains and curves all good, quite a bit of mains harmonics pickup with eqs engaged, somewhat mitigated by shielding.

Am futzing about now to see if I can improve the situation.
Also am comparing with my other unit to ensure there are no one off effects.
Will be powering that one using the first unit's PSU and located several feet away to check effect of onboard psu. If that helps, I'll use an external psu for both.
If necessary, I'll build up the other unit with onboard pots and switches and check for wiring noise.

I note that the wiring of the pots and switches would probably worsen the susceptibility to induced noise somewhat as compared to building it the way it was intended - with PCB mounted switches and pots.

Haven't done any listening yet.

All in all, an interesting project for a relatively low cost. Hopefully I can improve it further.

 

[alexc]

OK - I did the remote PSU test. External about 1m or so distant.

Nearly all of the EQ engaged,  induced noise gone 
Both channels virtually identical.

At max (12dB)  boost/cut of the bass band, noise floor goes from -82.5dBu now to around -77dBu
which is very respectable indeed. Mid band is -80dBu and hi band is -80dBu.

So - now to put the PSU in it's own external enclosure. Good thing I have 2 of these units to power.
Makes it a little more worthwhile. This is the first time I've needed to do this.
I still need to test external PSU with the unit in proximity to other powered up units. Could still be a problem.

Anyway, I have a go path now. Then to the listening!

Cheers

 

[desol]

Nice Alexc. Thanks for the post!

 

[peterc]

Good tests to know about.

Looking good....

 

[alexc]

Some more info regarding my specific case of rack mounting with wired pots and switches ..

Tested my unit with remote PSU then bringing a different powered up rack towards the unit under test - the noise floor gets progressively worse. With the other powered up rack on top or below,
the effect is the same as having an internal psu.

So, external psu only works with low noise if the unit remains well away (>60cm or so) from any psu.

---

Next test was to replace the inductors with some alternatives. I happened to have some old Collins shielded toroidal models at a couple of inductances.

Replacing the 1H inductor with a 720mH Collins, in the position closest to the psu, I get an improvement in noise floor at max low boost/cut from -42dBu to -68dBu BUT it remains very position dependent. (curve remains very similar!). Same story with the mid inductor, except I'll have to jigger the R-C values to accomodate my available inductor value of 470mH.

Being physically smaller, I can move it around in a way I cannot with the stock inductors.
So there is an onboard position which gives me a good compromise.

So, my plan is to go that way as I can't really locate these sufficiently distant from any psu in my studio.

None of this is particularly surprising - iron cored unshielded inductors with lots of turns are susceptible to induced noise at mains frequencies and harmonics from nearby power transformers.
Ferrite cores less so. Shielding helps a lot. Distance from source helps most.

I surmise that using the stock arrangement, within a populated 500 series rack, particularly with an external PSU would show acceptable performance if not too close to other racks.

 

[shabtek]

nice report

-- I wish I would have built a central outboard power supply at the start--could have saved much shoe-horning and cobbleing, and made for quieter gear

 

[alexc]

I finally settled the positions of the alternative inductors on the pcbs.

Very, very, VERY position dependent. 

With the onboard PSU I now am getting a flat eq noise floor of -82.0dBu and max boost -72dBu for the low, -79dBu for the mid and -80dBu for the high. cut figures are -79dBu, -81dBu, -81dBu respectively.  (eq hard bypass is -82.5dBu)

Both channels are within 2dBu of each other.

Which is great.  

Now mounting them permanently and then changing the freqs as desired and then listening.

Then, same again for the B unit.

Cheers

 

[alexc]

I completed each of the units with some adjustment of the centre freqs and bandwidths of the R/Cs for each band, I closed them up having found the curves as expected and now really good noise floors for all the bands. 

Put in my rack and checked everything was still fine (which it was) and settled in for a good listen.

It's all very nice.
Simple to use 
Sounds great on the lows, mids *and* highs !

Surely worth making especially with some good shielded inductors, perhaps a suitable Sowter or some of the custom ones in the White Market.

I have one after a dual1272 style pre and before a dual 1176 style comp
The other is after a dual api style  pre and before a dual what comp (in progress still)

Feeling well pleased wit them.  8)

Here's a pic of them. The great thing is with the relay bypass on each of the units, I can take the direct line out from my console to each of the preamps or eqs or compressors without necessarily passing thru each unit.

Or  opt for the full 'everything' experience of around 6 signal traffo and lots of discrete amps in each chain!

Life is indeed sweet.


Thanks again PeterC

Cheers

 

[benlindell]

Hey guys,

I was measuring a bunch of my gear this evening and when I got to my 5003s I saw something weird, there's a very pronounced HF peak that's always present, bypassed, in @ 0, any cutting or boosting. it's much above human hearing but you can see it reaches down and gives almost +3db@ 20k. Anyone have ideas as to where this comes from? Is it the transformer? Opamps?

 

[peterc]

Ben

I used a smaller cap in the 2520 feedback loop,  47p I think, whereas the original was 180pF. Could be there?

Peter

 

[jensenmann]

Looks more like a transformer resonance. Put a squarewave at 1kHz through the EQ and watch the output on a scope for ringing. If it does ring then add a Zobel network.

 

[jrowell]

Yes, the supplied output transformers ring like a church bell into 100K or 10K.

I think they are worse than an unloaded UREI 1176F output. If that were possible.

They should quiet down into 600 ohm.

But,....  600 ohm is too low. The -1dB point is at 8.5KHz. Something amiss there. More later, if it ends up being something other than just the TX.

A zobel would work, but must be designed. The better question is why?

I don't doubt API EQ out TX's ring without a 600 ohm load, because they never bothered with any comp or zobel, but to not drive 600 with full freq resp, ?

First though would be output amp comp, but Peter has mentioned the 180pF as being original to a 2520 amp.

A quick dirty fix is 1500 ohms, if going into 100K or 1600 ohm if going into 10K.

Pretty well maximally flat response, -1dB is 25 or 30K No ringing.

You could try 1.8K to 2.2K if you want a bit of top lift.

Also 1.5 or 1.6K happens to make the through gain almost exactly unity. Instead of up 0.75dB. Drop when EQ is engaged is about a 1/4dB, true with any loading.

They sound scary without the right loading.

James

 

[dmp]

I saw this HF behavior on a pair with the kit transformers, but not on a pair with EA transformers (from classic api). I don't know where the kit ones came from or why they had the HF lift, but it wasn't terribly noticeable in listening to them side by side.

 

[desol]

Peter mentioned this ages ago and provided samples. Fwiw, on the vocal sample...i thought it sounded nice.

All depends on what you like.

 

[outoftune]

Has anyone tried swapping cap values to get different frequencies? I'm considering changing out of of the LF caps so I can have a 60hz frequency.

 

[alexc]

Yes - I moved the centre freqs around somewhat by changing the R/C values.

You can calculate the changes or simply use a resistance/capacitance substitution box and check the results on a spectrum analyser application. You can see if your changes have introduced problems.