New 'ultimate' SSL buss comp clone ;-)

[nielsk]

I have had this problem with these units, get a power supply transformer that has more than enough current at as close to the regulation voltage (maybe 2 volts on the high side) or use some 10-25 watt resistors to bring it down to the right voltage, and check the voltage into the 12v regs as well. In most of mine 100 has worked better than 300 ohms... I have not figured out why, but if the regs have to work too hard sometimes they go a bit nuts. Remember that with the additional op amps you need more current

 

[nielsk]

Oh, and you meant 7815, not 7805....right?

 

[Jonkan]

yes 7815 of course... typo!
Ill try to lower the resistor value going to the 12v regulators a bit.

I have noticed one more thing. It seems that the strange 78L12 regulator works better when I disconnect the VCA control voltage link between the two pcbs. Then i got +-13v on the 12v rails.

When they are connected the voltage drop between the rectified voltage and the 270r resistor is about 13V, whereas on the other three its about 6v. What could be the cause of increased current draw on that particular regulator? I have searched for shorts on the pcb, but i cant find any.
/J

 

[SSLtech]

[quote author="Jonkan"]yes 7815 of course... typo!
Ill try to lower the resistor value going to the 12v regulators a bit.[/quote]...Raise, -right? -Not lower???

Keith

 

[Jonkan]

Never mind me, I was just chasing ghosts over here.

Turned out that the reason that i was getting so strange voltages, was that my multimeters battery was dying.


sigh...sometimes it takes a long time to realise the obvious.


Well now that thats over with, lets get on with this building stuff.

/J

 

[mhelin]

[quote author="SSLtech"]
M = L+R.
S = L-R.

The GSSL sidechain sees the two added together, therefore since:

M+S = (L+R) + (L-R),

and since L+R+L-R = 2L

you're left with 2L (left channel, plus 6dB) in the sidechain. Signals on the right channel would NOT do anything to compress the signal.

The ultimate "greater-of-two" sidechain would process both signals without preference or summing, and only the single-pole DC rectification is added, so it WOULD be able to do M/S.

The GSSL really wouldn't be usable for that purpose.

Keith[/quote]

Oh yes, stupid me. Anyway, because of the panning law used in mixers the virtual mid channel isn't actually the sum of L+R but 3-6 dB less, isn't it? However, stereo recordings made using two mics don't care about panning laws, or any other laws whatsoever.

 

[SSLtech]

[quote author="mhelin"]...because of the panning law used in mixers the virtual mid channel isn't actually the sum of L+R but 3-6 dB less, isn't it? However, stereo recordings made using two mics don't care about panning laws, or any other laws whatsoever.[/quote]
often -4.5dB, but certainly usually between -3dB and -6dB.

However this doesn't matter, and ONLY affects "moving-pan" images where the fader stays still. -If you're bringing a kick drum up the middle, it doesn't matter whether the pan law is -3dB at center or -6dB at center... you're going to place the fader knob somewhere within a 3dB-wide range to produce EXACTLY the same result, so typically the fader will just be placed 3dB higher if the law is -6dB than if the law was -3dB, and the pan law immediately becomes stunningly irrelevant.

Now, regarding the stereo miking bit:

For Two-mic-stereo techniques, a center signal in an anechoic room will usually appear in equal amounts in both microphones, but with varying levels of gain compared to 'wide' position gain. The precise level difference varies with ORTF, X/Y, M/S and spaced-omni, and there is no single concrete number. Note I say anechoic, because (non-phase-coherent) reflections will affect apparent level measurements while not adding and canceling sufficiently simply to refer to with a single number. -In addition, techniques like spaced-omni will provide lots of timing difference (therefore comparatively poor cancelation or summing) but not much level change; relying on the arrival times to suggest spatial placement rather more than coincident techniques.

-Either way, there are times when one may be perceived to work distinctly 'better' than the other, and a switchable sidechain is a massively important tool.

[quote author="mhelin"]However, stereo recordings made using two mics don't care about panning laws, or any other laws whatsoever.[/quote]

Well, they DO care about the laws of physics... and I'm not aware that anyone's planning to repeal any of those laws any time soon! :wink:

Keith

 

[Jonkan]

Done! Well almost. I need to attach the toroid properly before closing the lid, and the panel is just a prototype that i made until i get my engraved panel.

It sounds great!!!

I have one strange issue though. I measured the box using rmaa, and it seems that i have alot more THD in the unit when its bypassed. Everything else seems just dandy.

Not that i can expect too accurate values measuring this way, but there seems to be alot of difference in distiorsion when it is bypassed and not, even to my ears.

The values reported where: 0.017% THD when bypassed, and 0.0079 when not. I didnt measure at which level this was at exactly coming out of the soundcard, but the levels where matched between the two tests.

It confirmed what i was hearing before i did the test: Some distorsion in the sound when the compressor is bypassed.

Is this normal? And if not, any ideas where to start looking for errors?

/J

 

[Jonkan]

Did i kill this thread? Im sorry if im asking stupid questions. Im am slowly learning, but sometimes i build stuff alot faster than i learn what the hell i am doing.. If you know what i mean.
:green:

Heres a picture of the THD test in RMAA of L+R channels in my compressor. The compressor was not in bypass mode, no compression, or makeupgain applied.

As you can see the values of distorsion differ alot from the two channels, and im trying to figure out why?

Could this be normal, and be because of variations in the pretrimmed 2180 vca:s for instance?

As everything seems to be working fine otherwise i think i have connected everything correctly. The only other thing that differs between the two channels that i can think of is that the master has to drive the cheap behringer meter, where the slave doesnt. But since the signal is already rectified and buffered, this shouldnt make a big difference in THD right?

If you have any ideas on where to start looking for errors, feel free to tell me. Maybe i can trace the signal on my scope until i find the culprit?

anyway, heres the pic:

/Jonas

 

[Rob Flinn]

Jonas,

Maybe this is a waste of time but have you done the same test with the channels to your sound card swapped to the channels of the compressor ? At least doing this you will know that it is definitely the compressor.

 

[Harpo]

[quote author="Jonkan"]Heres a picture of the THD test in RMAA of L+R channels in my compressor. The compressor was not in bypass mode, no compression, or makeupgain applied.

As you can see the values of distorsion differ alot from the two channels, and im trying to figure out why? [/quote]
Seems supply related. See the peaks at 50,100,150,200...
Tightly twist the leads between transformer secondary and rectifier should help it.
-Harpo

 

[SSLtech]

Is that measured with the lid ON or lid OFF?

Looks like it could be dominantly field pickup to me...

Keith

 

[Jonkan]

Thanks for the suggestions. Its not the soundcard, i tested a straight loopback through it and the results were almost exactly the same for both channels. About 0.0007%

The measurement in the pic is actually with the lid off, but the results are the same with lid on, i tried that too.

Ill try twisting the wires and see if that helps or not.

Any other ideas?

/J

 

[keefaz]

[ot] Where did you get the lorlin power switch? :wink:

 

[Jonkan]

I had it at home actually. But you can get it at RS components, and at ELFA, but they are more expensive.

/J

 

[Jonkan]

I havent really fixed all the problems yet with my unit. But i have done some listening tests to some mixes. I put on a friends recording and sent it to the compressor. Even though this is mastered material i felt that it could use some tender compression on the mixbuss.

Result=It sounds really good!!!!

On some mixes where i felt the snare was sticking out a bit, and things werent "gel:ing" the way i wanted to, i set it to oxford mode 2:1 ratio, slow attack, auto release, about 0.5-2 dB reduction at the most and that really made the whole mix sound alot more like a record.

I compared it to the gyraf mode and like i read, its definately different. The Oxford works better for me on 2bus so far, because it sounds more "grounded" in the center. Kickdrum, snare and bass seem to have more impact to my ears. Not an earthshattering difference though, but still there.

This thing is going on my 2bus as soon as i can fix the minor issues with it.

Just wanted to give everybody a quick review. I can definately recommend building this version.
/J

 

[Jonkan]

I think i might have found the culprit!

When i touch/move the wires going to the bypass switch the THD value for both channels become almost the same and not like before where i had wildly different values.

Sadly both channels THD increases to about 0.020% instead of 0.007% like i had before on one of the channels. Strange indeed.

Also sometimes the makeup gain stops working, which also leads me to think it is the switch that is faulty.

I sure hope that is it anyway. By now i have double and triple and quadruple checked everything. Its getting a bit tiring, to say the least.

/J

 

[Dr nEon]

Jonkan,

I dunno if this helps , but looking at the pic u posted, it looks like u opted for the bypass switch to defeat makeup gain when bypassed. I think i see five wires leading from it...correct me if I'm wrong.

If so , then the defeat switch connection should be linked to the slave switch pcb , and I can't see that happening on your pic , unless u soldered it on the underside of the board.

I am just assembling my own attempt at the ugssl, and I'm thinking of having the makeup gain always active , which leads me to ask how should this way be wired?

The instructions on page 10 mention linking that point if u wanna defeat m/up in bypass , but surely that point needs to be linked regardless ??

Cheers!

nEon

 

[Sleeper]

Hi Neon,
I'm sure it's in the big ssl post.
This is a confusing point of the build for sure. I'd simply answer you, but I just picked up my ultimate ssl after letting it sit for a few busy months.
That's also one of the things I need to look at.

Hope I'll be testing really soon. I've got a couple more issues to deal with and I'll be there.

It would be great if somebody could check my head here on the following sidechain summing/filter issue. I keep reading speculation about this but the posts seem mostly confused.

from what I can tell no changes are required.

I'm using Gregs sidechain filter board X2
the resistors into this are 20k instead of 47k(this seems to be the source of the confusion... it was for me)
Nextthe sidechain is buffered by a 5532
then the filter and thrust circuitry
Next a 5532 output buffer with trimmable gain and 47k resistors on the output.

Looks to me like the sidechain should be very happy with this.

as it's been said earlier in the thread that I probably won't be able to pull this out of the rack once it's placed (and I don't have a distortion analyzer) I just want it to be as right as possible.

Thanks
Kelly

 

[Jonkan]

Neon, thanks for your suggestion.

I thought about the same thing earlier, but then i realised that the makeup gain is already linked through the "F" points.

My unit is working correctly, but one side has alot more noise and THD. Well not alot, but i can measure it, which makes me want to fix it, since im such a nitpick.

I even tried swapping the pcbs so that the slave channel, which measures fine, would be the master (of course moving the 3m3, and 2k resistors that separate the L and R channels)

The problem seems to always be with the master channel, no matter what i do. The pcb that is the slave always measures fine, and the one that is master always measures bad. So both pcbs seems to be working correctly on their own, or as the slave.

THD+noise is reported at 0.015 respective 0.050 for the other channel by RMAA, and most of it is around 50,100,200, etc, HZ which leads me to think about psu problems. But since both channels excitbit these problems , but only when being the master im not sure where to start looking for faults.

Im actually thinking of splitting these pcbs to two regular GSSl comps and just wait for Keith to finish his Ultimate Gssl design PCBs, which i hope will be coming pretty soon.

/J