GSSL VCA comparisons

[[silent:arts]]

I have done some measuring (Audio Precision System Two cascade) between different VCAs (sidechain VCA always 2150A), all in the same GSSL board. There are two VCAs left to measure (dbx202 (thanks to Jens) and THAT2002T), I will update this first post when I'm done with it.
The results are surprising (in my opinion):

the board itself measured:
THD&N @ 0dBu: 0,0012%
Noise (left / ride): -94 dBu / -97 dBu
Dynamic Range: 110,19 dB

dbx 1252 VCAs:
THD&N @ 0dBu: 0,008%
THD&N @ 0dBu with 10dB MakeUp Gain: 0,022%
Noise (left / ride): -80 dBu / -84 dBu
Dynamic Range: 94,02 dB

THAT2159 VCAs:
THD&N @ 0dBu: 0,006%
THD&N @ 0dBu with 10dB MakeUp Gain: 0,012%
Noise (left / ride): -80 dBu / -83 dBu
Dynamic Range: 94,12 dB

THAT 2150A VCAs:
THD&N @ 0dBu: 0,005%
THD&N @ 0dBu with 10dB MakeUp Gain: 0,011%
Noise (left / ride): -80 dBu / -83 dBu
Dynamic Range: 93,85 dB

THAT 2180LB VCAs:
THD&N @ 0dBu: 0,023%
THD&N @ 0dBu with 10dB MakeUp Gain: 0,023%
Noise (left / ride): -80 dBu / -83 dBu
Dynamic Range: 93,86 dB

THAT 2181LB VCAs:
THD&N @ 0dBu: 0,055%
THD&N @ 0dBu with 10dB MakeUp Gain: 0,076%
Noise (left / ride): -80 dBu / -84 dBu
Dynamic Range: 93,84

 

[ReggieReddog]

Alright, I picked the worst one (2181LB)!
I'll just pretend that extra THD is the good-sounding kind of distortion :razz:

 

[[silent:arts]]

[quote author="Viitalahde"]Come one, 0.023% vs. 0.006% - you won't hear the difference. The numbers are just good for wankers like me.[/quote]
... and it's good for spending EUR 50.- for a Be*ringer Cybermix on eBay for 32 x THAT2159 :green: ...

 

[radiance]

I would like to know how the dbx202 behaves....

 

[[silent:arts]]

[quote author="radiance"]I would like to know how the dbx202 behaves....[/quote]
will come ... just haven't had enough time yesterday

 

[smallbutfine]

What a positive surprise!
Yep, seems to be a good choice, the old cybermex...i hope most of my 2159 survived the extraction :twisted:
...maybe i try out in my 3630 to be sure.

Very nice idea to test them all. :thumb:

Kind regards

Martin

 

[[silent:arts]]

the first one I used a desolderingstation, but was killing a few pins. the last one I just used a heatgung :twisted:
- and everything (incl. the TL074s) survived. :green:

 

[gyraf]

Yep, heatgun is the way for bulk-disassembling pcb's. But take care - the fumes are quite unhealthy. Better do this outdoor..

 

[chrissugar]

By the way the value of the THD is not too relevant. You can have two devices that have 0.01%THD and sound totally different. It is the harmonic structure (order and level of each harmonic) of the distortion that is important and will have impact on the sound of the device.

chrissugar

 

[gkhebert]

silent:arts,

It seems that neither the 2180B nor the 2181B are meeting their specifications, which is something that we take very seriously at THAT Corp. I'd like to ask a few questions to find out why.

1. Did you trim the symmetry for the 2181B? They should typically trim to less than .003% with the with 0 dBu, 1 kHz input (presumably below the compressor's threshold).

2. For the 2180LB, was the resistor between pins 2 and 4 for the 2159/1252 left in the PCB when this was measured? The 218x VCAs have an internal 25 ohm resistor across these pins. Paralleling it with the external 51 ohm resistor will upset the internal symmetry trim. (BTW, you can trim the 2180 devices if you so desire.)

Some additional information: The THAT2159 and uPC1252 are the same device, with the same specs. The uPC1252 was sold to dbx licensees for use in licensed noise reduction products. (Where do you guys find these things?)

You should find that the 218x-series VCAs will have substantially lower THD as you increase the measurement frequency. At the levels that you used, the 218x should essentially have flat THD in the audio band at unity gain (assuming no capacitive crosstalk of signal into the control port). The 215x parts will show increasing THD above 2 kHz (and this will, of course, get worse at gains higher than unity due to gain-bandwidth effects.) The 215x devices will also exhibit higher THD at gains other than unity.

Regards,
Gary Hebert

 

[smallbutfine]

@Jaako:

How do you remove these things? I use flux, extra solder and tons of clean solder wick and I still manage to snap a pin or two.

Exactly the way i killed the first two vca's...
The b* boxes are really not easy to desolder (it's pretty much more easy to throw them away...)...
i used a 10? 1500W heat gun from the supermarket (i just bought it out of frustration about the desoldering disaster here!) and had to go to the highest temp: 500 degree celsius at 480l/min.
i used the smallest point application and let the pcb stand vertically.
i moved the heatpoint a few times over the solder points of the solderside and pulled the ic's out with pliers as soon as the solder went shiny and tried to use as little force as possible until the pins went out pretty much automatically.
I went over to the next one of this pcb area because i didn't want to reheat the same components twice.
From time to time i knocked at the pcb and mostly all components fell out.
This procedure was very fast.
Be sure to not overheat (pcb turning brown...).
Be sure to get your lungs away from this!!! You can seriously etch yourself into a bad health damage with a strong breath, that's absolutely NOT funny. Despite i took very much care of it at fresh air i managed to get a very bad cough for some days.
The smaller temp of my heatgun (300 degree celsius) didn't manage to get out ANYTHING out of this pcb.
Try out the right technique on some unimportant components before going to the vca's.
You should manage to get out > 20 vca's in a very good optical condition.
I would not recommed cooling down the parts somehow afterwards...the resulting tension between the different materials cooling down could break them more easily than the high temp.

Kind regards

Martin

 

[[silent:arts]]

[quote author="gkhebert"]It seems that neither the 2180B nor the 2181B are meeting their specifications, which is something that we take very seriously at THAT Corp.[/quote]

Hi Gary,
THAT's nice :wink:. I think the main reason for this result is the schematic they were tested in. the GSSL is a nice fine project, but NOT designed to be used with the 218X series (GSSL schematic).

to answer your questions:

1. yes, I trimed the symetry of the 2181B, and the best result (0,055) was at the full CCW of the trim pot. but again, the dist. trim circuit os not the one you describe in your datasheets. measuring was done below threshold.

2. regarding to the shematic of the GSSL I left out the 68R resistor from pin 4 to GND and the 10K resistor from pin 4 to the trim circuit.

you can find PC1252 VCAs in older, THAT2159 in newer Behringer gear.

I'm sure the 218X measure much better with a modified circuit of the GSSL :thumb:

regards
Volker

 

[BR]

Hey Silent,
thanks for that...
are you thinking about the dbx2001 also??? I have built a few with these and would love to know how they compare to the THAT stuff.

Hey Gary,
thanks for posting .... we like your stuff. .. a lot.. :thumb:

 

[ReggieReddog]

I'm so confused about the proper way to implement the 2181LB in the GSSL. :sad:
This thread doesn't seem to really clear anything up: http://www.groupdiy.com/index.php?topic=358

I thought I was supposed to leave out the trim pot, so I did. And I didn't know I was supposed to remove the 68R and the 10K, so I didn't. Is there just not a clearcut way to do this?
Hmph.

 

[[silent:arts]]

[quote author="BR"]are you thinking about the dbx2001 also???[/quote]
well, I have no dbx2001 - but if you send one or two to me I can do the same measurings.

ReggieReddog, you leave the trim pot etc out for the 2180. for the 2181 you should leave everything in. and if you like the sound of your unit don't think about a proper way.

I haven't done this measuring session to find the best VCA or something like that, I just wanted to have a look how different VCAs react to this given circuit. and none of the candidates measure bad, btw.

 

[gkhebert]

Hi Volker,

After looking at the schematic, I can see that the trim range would be inadequate for the 2181LB. With the 1M resistor in series with the wiper of the pot, the maximum adjustment at the symmetry pin is about +/-375uV (due to the 25 ohm internal resistor). The 2181LB requires +/-1.5 mV, worst case. Changing the 1M to 240k will fix this.

With the 2159/1252, the 68 ohm external resistor allows a trim range of +/-1 mV, which is enough for most (but not all) devices They require +/-2.5mV worst case (a 390k resistor replacing the 1M would cover this).

Your approach to the pre-trimmed 2180LB seems correct. The THD seems a bit high, but the 15k input and output resistors (they're tested and trimmed with 20k) may account for this. I still think that there may be something else going on.

One other comment:

The 10k resistor between pins 3 and 5 is really unneccesary. It was an attempt to minimze control voltage feedthrough in the 2159 that wasn't very effective. It's totally unnecessary for the 218x devices.

Reggie,

The substitution of 218x devices for 215x devices is covered in a design note on our website. Go to www.thatcorp.com and click on Design Notes. Look at DN137.

Regards,
Gary

 

[ReggieReddog]

Hey, really cool of you to get involved and help us out here Gary.
Looks like I have some good info here; now I just have to make my peabrain understand it. Although, I thought the compressor still sounded quite great even with the "off" parts in the circuit, so it may not get much better than great. :grin:

 

[[silent:arts]]

Gary, THAT's great again :wink: :thumb:

I stuff an other board with your suggestions for the 2181, and will measure again. I will have another look for the 2180s too.

the time I started doing my GSSL it was impossible for me to get 218x VCAs, that was the reason I bought the Behringer Cybermix on eBay. today I know Cedos and can by THAT ICs in single quantities :thumb:
would be nice to see you having a few small dealers like Cedos over the world, because for things like the GSSL the question about the VCAs is NOT which model you want, it is which you can get.

when I have finished all the tests I will draw a schematic and component overlay with all the changes for the different VCAs.

thanks again to Gary for visiting and supporting us, and to Frank Thomas who gave me all informations for the 1252 and 2159 VCAs. THAT is a nice company. :thumb:

 

[[silent:arts]]

roger,

I can do measurings for the tracking, too. best way for me was to use a trim pot (see http://www.groupdiy.com/index.php?topic=8897)

threshold is another thing, but I don't think it has anything to do with the VCAs.

 

[martthie_08]

A little while back, I did a web page showing (among other) the exact modifications needed for the current production THAT 2181 VCA's for the audio path section of the GSSL, based on the document DN137 from THAT.

If someone has some webspace, I will gladly put this up. I also made a couple of the changes suggested on this board, this really changed the compressor to the better, especially in the high frequencies.

cheers, Marten