SB4000 Support Thread

[MarioXLR]

I'm really curious about this!!!

King regards alllllllll

Hello, this is my first post. I've been enjoying my SB4000 which I built successfully thanks to the many great comments on this thread. One thing I'm very curious to try is to replace the TL074's with OPA404's. Would I need to change anything else, or is it just a direct swap? See below for why I'm so curious to try it:

...I'm curious if I should upgrade the TL074's with OPA404's that I have lying around:

Had a cool experience with OPA404 I did not expect.

I replaced a TL074's of a GSSL (also with dual or turbo or oxford mode) sidechain rectification/ratio amp with OPA404. Gain reduction did not change as expected, but the sound of the compressor transformed dramatically! With OPA404 I could do much deeper gain reduction with less artifacts. It removed the "phat" SSL trademark sound, and it sounded more like some kind of transparent mastering unit. One friend commented that it "sounds like a plugin now". I guess TL074 has some serious issues with transients. Looks like speed counts immensely when we are buffering rectified (sharp discontinuities) audio. Looks like the slow TL074 "distortion" is a big part of the SSL sound.

 

[friesdan]

Hi Mike,
A little confusion in the documentation here.
On the SB4000 main BOM, I see R76 and R96 listed two times, with 2 different values.
Both on page 2, 5th line down they are listed as 220K (as they also are on the Quad VCA BOM, to be replaced with 56K resistors), and then 11th line down, they are listed as 680K (as they are also printed on the PCB).
I'm using the Quad VCA, so I'm planning to replace them with the 56K resistors, but there are a whole bunch of resistors listed on the PCB, so I'm a little nervous about soldering these in on top of a printed value which doesn't match the Quad BOM's instructions.

Also on the main BOM, 5th line down, in the notes section, R127 is listed as needing a different value (along with R96) for alternate VCAs - and this is not on the Quad VCA board. I'd like to be sure I'll be using the original value resistor for this one.

Thanks for your help and your time.
-Dan

 

[MountCyanide]

Hey there--
For R76/96, have a look at the schematic. It's got a chart for these differing resistor values dependent on your chipset. If you had a hard time finding the schematic like I did, a google search led me to find a link to it at diyrecordingequipment.com

 

[friesdan]

Thanks for your help MountCyanide, it was just what I needed.

I have another quick question for anyone with an experienced opinion:

A few resistors were missing from my kit, (22k x2, and 12k) - and I have some carbon film resistors with the right values on hand (although tolerance is 5% as opposed to 1%).
OK to use these carbon films in place of the metal film ones? Or should I have some patience and just order the missing ones in metal film?

My vague understanding is that metal film are superior in terms of clean and stable.. but is it enough to matter with these 3 resistors?
Thanks,
Dan

 

[Harpo]

A few resistors were missing from my kit, (22k x2, and 12k) - and I have some carbon film resistors with the right values on hand (although tolerance is 5% as opposed to 1%).
OK to use these carbon films in place of the metal film ones? Or should I have some patience and just order the missing ones in metal film?

Be more specific. IIRC there are 8 resistors (R24,R28,R29,R45 and R32,R35,R36,R37) with nominal value of 22K at the balanced line receivers. For best CMRR you want these close matched for same value, not absolute value. The 1% tolerance 22K might have a real world value somewhere in between 21,780R and 22,220R, the 5% tolerance 22K might have a real world value somewhere in between 20,900R and 23,100R when you measure these with your multimeter to sort for best matches. If your 5% tolerance parts have the same or closer readout than the 1% tolerance parts, they would be the better fit. (Your goal might be a better than 0.1% tolerance match).
For the 12K value (R40 and R42), parts value is not critical.

 

[MLAudio]

Just ordered a kit. Spent the last two evenings reading all 62 pages of this thread. Can someone point me to a schematic? I looked at the first post and didn't see one. Thanks.

Oops, nevermind. I found one at diyrecordingequipment.com.

 

[EmptyMusicBox1]

hello !
i have just a question about TilT position on the SB4000

It's a HPF and Boost to 8khz on the Sidechain circuit ?

thanks

 

[benranco]

I've been working on getting my THD as low as possible, and one thing I've noticed is that it seems to fluctuate over time even if I don't touch the trim pots. I've adjusted VR3, VR4 and the 50K pots on the quad vca boards using a 1kHz test tone and watching the node at 2kHz, following the procedure ruckus suggested. With the test tone peaking at just below 0dB, I calibrated it so the node at 2kHz peaks at around -54dB. However, if I leave it on for a few hours, the amplitude of the 2kHz node tends to grow and shrink slowly over time. I've seen it get as high as -44dB, although it usually stays between -48dB and -52dB. Is this normal behaviour? Is this due to fluctuations in the electrical current, or something internal to the SB4000? Thanks.

 

[Harpo]

With the test tone peaking at just below 0dB,...

I could be wrong, but your 0dBwhatever seems to be 0dBFS that has no relation to the analoge world without reference. Depending on your converter, 0dBV (=1V RMS) might be in range of -12 ... -18dBFS. Set your test tone for a 1V RMS readout between XLR-input pins2/3 to get the dBFS relation for your converter. (just for a perspective, if this would be -18dBFS, your 0dBFS test tone had a 7.94VAC=11.23Vpk=22.47Vppk level).

I calibrated it so the node at 2kHz peaks at around -54dB.

=0.2% or factor 0.002 below (probably too high) input level.

However, if I leave it on for a few hours, the amplitude of the 2kHz node tends to grow and shrink slowly over time. I've seen it get as high as -44dB, although it usually stays between -48dB and -52dB. Is this normal behaviour? Is this due to fluctuations in the electrical current, or something internal to the SB4000?

Fluctuation might be temperature related as one possible source. There are a lot other (any incoming handy calls, ... ?)
The VCAs where set for 0.00V control voltage?
Bandwidth of your measurement was limited within 10Hz...20kHz?
The case was shut to keep RFI/induced/radiated garbage out?

 

[benranco]

Thanks Harpo. Yes I was using 0dBFS, thinking that with a hotter signal the node at 2kHz would be bigger, and hence easier to adjust. Is it better just to calibrate with the test tone set at 0dBV? I've adjusted my test tone to read 0dB on another compressor's VU meter. So now a -15.7dBFS test tone results in a 2kHz node that has over the past few hours fluctuated slowly between around -62dBFS and -56dBFS. There was an initial drop as the unit warmed up after first turning it on, but since then it has gradually crept up and maybe gone down slightly a few times. It does seem to have a ceiling which it doesn't go beyond though, and the change seems to be reflected equally on both channels. Is this normal or something to be concerned about? The compressor sounds great, for what it's worth.

Fluctuation might be temperature related as one possible source. There are a lot other (any incoming handy calls, ... ?)
The VCAs where set for 0.00V control voltage?
Bandwidth of your measurement was limited within 10Hz...20kHz?
The case was shut to keep RFI/induced/radiated garbage out?

The case is shut. I'm not sure what you mean by VCAs set for 0.00V control voltage, my knowledge isn't very deep. I followed the build guide instructions for control voltage calibration, and adjusted VR8 and VR7 until the AC voltage at TP11 and TP12 matched the voltage TP8 and TP10, respectively. My multimeter is pretty cheap, so I can only guarantee accuracy to one decimal place though.

I'm not sure what you mean by the bandwidth of my measurement. I'm just passing at 1kHz test tone through the compressor and observing the returned signal through a frequency analyzer. I did also do a THD frequency sweep test between 100Hz and 10kHz using RMAA, and the THD level at -15dbFS was around -78dBFS, while the THD level at -6dBFS was around -84dBFS. I found it odd that the THD level was lower at the higher test level. Perhaps because I calibrated it with an extremely hot input signal?

 

[friesdan]

Here's a warning --- if you want the molex to work, you're going to have to buy a $300 crimper. I'm kicking myself now for having soldered the molex connectors to the board.
Buy a desoldering tool?
Or just solder direct to the little molex pins and heatshrink?

After 2 days of trying this with needle nose pliers, and a "regular" crimper, I am very frustrated.

If you buy one of Mikes kits and aren't already a molex guru, do yourself a favor and stick to the green connectors on the bom... Save a lot of trouble that way.

 

[Harpo]

...The compressor sounds great, for what it's worth.

That is the most important part of it.

I'm not sure what you mean by VCAs set for 0.00V control voltage, my knowledge isn't very deep. I followed the build guide instructions for control voltage calibration, and adjusted VR8 and VR7 until the AC voltage at TP11 and TP12 matched the voltage TP8 and TP10, respectively. My multimeter is pretty cheap, so I can only guarantee accuracy to one decimal place though.

For THD adjustment you set the audio-VCAs Ec- port (more exactly the differential between Ec+ and Ec- ports) to 0V for the VCAs current-in=current-out bypass condition with a moderate (maybe 0dBV) 1kHz sine test tone running thru your unit. This 0V at the Ec- port is either done by temporarily shorting out the 120R shunt arm resistors R83 and R125 or by setting the unit for 'external key' with external key input shorted to gnd (FI link T with S inside a TRS plug, ...) so the one side connected caps C23 and C24 don't pollute your measurement.

I'm not sure what you mean by the bandwidth of my measurement. I'm just passing at 1kHz test tone through the compressor and observing the returned signal through a frequency analyzer. I did also do a THD frequency sweep test between 100Hz and 10kHz using RMAA, and the THD level at -15dbFS was around -78dBFS, while the THD level at -6dBFS was around -84dBFS. I found it odd that the THD level was lower at the higher test level. Perhaps because I calibrated it with an extremely hot input signal?

Double check the RMAA setting in /general/'Analyze noise and distortion only in 20Hz-20kHz range'  checkbox is enabled for the THD+N measurement, else the +N part might dominate your numbers.
Your varying results might be related to prementioned one side connected caps.

 

[benranco]

For THD adjustment you set the audio-VCAs Ec- port (more exactly the differential between Ec+ and Ec- ports) to 0V for the VCAs current-in=current-out bypass condition with a moderate (maybe 0dBV) 1kHz sine test tone running thru your unit. This 0V at the Ec- port is either done by temporarily shorting out the 120R shunt arm resistors R83 and R125 or by setting the unit for 'external key' with external key input shorted to gnd (FI link T with S inside a TRS plug, ...) so the one side connected caps C23 and C24 don't pollute your measurement.

Thanks again Harpo! I did what you suggested and was able to improve things with further tweaking using a more moderate (close to 0dBV) 1kHz tone. The 2kHz distortion node is now lower in amplitude and seems to change over a smaller range. I'm not sure if shorting the external key input reduced the amount of variation while testing, because it seems to be better now even when it isn't shorted. I've also noticed that the variation only seems to happen when the compressor is in external key mode. It appears stable when in internal key mode. Given that I almost never use the external key, I think I'm happy enough to give it a rest. The unit sounds amazing, and I'm happy with the THD levels I'm getting. Thanks!

 

[Harpo]

I'm not sure if shorting the external key input reduced the amount of variation while testing, because it seems to be better now even when it isn't shorted.

For normal operation when using an external key, there would be a low impedance source connected to this external key input, so these caps would no longer be only one side connected.
Placing maybe 47K resistors between input side of these caps and 0V reference voltage (maybe at the external key socket) should help for a permanent fix if this bothers you.

I've also noticed that the variation only seems to happen when the compressor is in external key mode.

This is depending on your (maybe noisy) external key source and we're talking fractions of millivolts here that might modulate the VCA ports.

 

[MLAudio]

This is only half of the story. What is your dBFS reading to get a +0dBV readout ? Might be in range -12....-18dBFS.

Yep - it's -13.8dBFS

I'd calibrate the sidechain section following signal and the schematic, so order will be different.

Initially power on the unit, engage compressor (disable the hard bypass), set the unit for no makeup gain, set HPFilters to off position and set ratio to 2:1.

For step 2 adjust audio path unity gain setting by trimmers VR1 and VR2 with unit set to soft bypass (engage External key with no signal (nada millivolt, niente dB, nuthin' whatever) on external sidechain input. Maybe short out tip and sleve at your external SC plug to ensure this). The 'with the unit in bypass' might be confused with the hard bypass relay that simply connects input to output.

For step 3 with 1kHz sine signal present, External key disabled and unit switched to default SSL-mode adjust VR8 for TP11 reading the same as TP8 and adjust VR7 for TP12 reading the same as TP10. Note these readings for the following step.

For step 5 (the 2nd. of the 3x step5), still in SSL mode, set HPFilter to T1 and adjust VR5 and VR10 for same reading as in previous noted readout when measuring at TP11 and TP12.
Set HPFilter to T2 and adjust VR6 and VR9 for same reading as in previous noted readout when measuring at TP11 and TP12.

For step 5 (the 1st.) Drop feeding signals by 10dB below 0dBV (0.316V AC, something in range -22...-28dBFS) both L/R reading the same level. Turn threshold full CCW (probably always compressing). Note the L/R meter readout of your return signal in your mixer, DAW or measure return signal AC voltage. Now increase feeding signals by 10dB to 0dBV. Tweak VR16 for best average for
+5dB more (or noted voltage times 1.778, probably about 0.562V AC) than previously noted return level in 2:1 ratio setting,
+2.5dB more (or noted voltage times 1.259, probably about 0.398V AC) than previously noted return level in 4:1 ratio setting,
+1dB more (or noted voltage times 1.122, probably about 0.355V AC) than previously noted return level in 10:1 ratio setting.

For step 4 increase threshold to about -10dB and adjust VR17 to reflect the GR meter showing the same amount of compression at differing ratios.

Hi Harpo,

Thanks again for your very detailed guide.  Hopefully it can assist someone else too because it was good to do everything in that order for me.

When I went through it just now, most of this process worked extremely well right until your last step!  (Step 4 in the guide).  Ratios are all looking good (measured with dBFS as well as double checking with AC) ... all of the various test points voltages match up ... but the compression meter still only gets as close as 12db when the compressor is knocking down 10db.  It is quite accurate at the end of the scale however - 20db on the meter equates to pretty much 20db of compression measured in my DAW.

Ah well!   I think it's just one of those things that may not work out perfectly.

Thanks again for your advice!

Same issue here. Just finished my build (Quad VCA). The only calibration step I couldn't get right. Followed all this to a tee. I'm using the same Hairball 1ma meter. VR17 is at it's CW max. Do you think a 10K pot there instead of the 5K would help? Using my ears, the unit sounds good.

Only other note is that I had to put some lower value fixed resistors on the PS board for VR18, VR19 because with the supplied 1.33K's my 15V was around 17.5V. I put something like a 1.2K in for them as I didn't have a 2k or 5k pot around and got it down to +15.32V and -15.35V. Does anyone see any issue with those voltages? If so, I'll order up some 2K pots.

 

[Harpo]

I'm using the same Hairball 1ma meter. VR17 is at it's CW max. Do you think a 10K pot there instead of the 5K would help?

I'd assume this "1mA meter" being an about 600uA meter for real. A 10K trimmer (or for an easier fix an about 3K3...4K7 fixed resistor at the meter terminal in series to the already populated 5K trimmer) should help.

Only other note is that I had to put some lower value fixed resistors on the PS board for VR18, VR19 because with the supplied 1.33K's my 15V was around 17.5V. I put something like a 1.2K in for them as I didn't have a 2k or 5k pot around and got it down to +15.32V and -15.35V. Does anyone see any issue with those voltages? If so, I'll order up some 2K pots.

Not only resistors come with tolerances and again only half of the story. If your R179 and R181 would be 120R for real instead of your lower resistance fitted values, the 1K33s for an about +/-15.17VDC setting would be close enough. You can keep these supply voltages at your +/-15.3VDC.
(With +/-17.5VDC supply rails you are close to boarderline for U14 and the 5K1 VCA bias current setting resistors R105 and R146 would better get increased.)

 

[MaNiAC]

Hi,
I just received the parts and PCB from Serpent Audio (Thx Ruckus!), and I'm gonna start building the unit, and I have a question:

I've built 3 GSSL a few years ago, 2 of them have oxford mode, one have the original GSSL mono sidechain. I don't really find a use for the GSSL mode (and I need  space on the frontplate for a future CnB pot) ...

So if I want to skip the Oxford/Gssl switch, I only have to:
not populate R140, R11, J10, K5, D5, C2, R12, change R13 for a 10R resistor and link point 4 to point 3 on the K5 slot?

Please, can somebody confirm i'm right?

If you think  it can be interesting to keep the GSSL mode, please let me know, too.

Thanks to Ruckus for this project,  Harpo and the others for their help and sharing their knowledge.

Niko

 

[MLAudio]

I'm using the same Hairball 1ma meter. VR17 is at it's CW max. Do you think a 10K pot there instead of the 5K would help?

I'd assume this "1mA meter" being an about 600uA meter for real. A 10K trimmer (or for an easier fix an about 3K3...4K7 fixed resistor at the meter terminal in series to the already populated 5K trimmer) should help.

Only other note is that I had to put some lower value fixed resistors on the PS board for VR18, VR19 because with the supplied 1.33K's my 15V was around 17.5V. I put something like a 1.2K in for them as I didn't have a 2k or 5k pot around and got it down to +15.32V and -15.35V. Does anyone see any issue with those voltages? If so, I'll order up some 2K pots.

Not only resistors come with tolerances and again only half of the story. If your R179 and R181 would be 120R for real instead of your lower resistance fitted values, the 1K33s for an about +/-15.17VDC setting would be close enough. You can keep these supply voltages at your +/-15.3VDC.
(With +/-17.5VDC supply rails you are close to boarderline for U14 and the 5K1 VCA bias current setting resistors R105 and R146 would better get increased.)

Vielen Dank, Harpo! I will keep my 15V rails and try the resistance value increase on the meter circuit. It makes sense on the meter since they are not precision instruments.

[UPDATE] Added a 5K resistor in series with VR17 and all is well in the land of gain reduction metering. For those who want or need this fix, here is what I did. I cut the trace on the back side of the PCB between C121 and VR17 Pin 1 (CCW) and then soldered the resistor on the backside between those points. This allowed me to correctly complete Calibration Step 4) Meter/Ratios with the 1ma (slightly less in reality) Hairball compression meter.
For those who want, the cleaner approach would be to replace VR17 with a 10k VR. An example would be Mouser: 81-PV36W103C01B00.

On the voltage rails, looking back, I would also use the optional 2K pots at VR18 and VR19 on the power supply board (only fixed 1.33K are supplied). This will save you some time in the long run. I would pre-trim them to about 1.2K and ensure that the rails are adjusted to +15/-15 VDC before connecting the main board.

 

[Harpo]

So if I want to skip the Oxford/Gssl switch, I only have to:
not populate R140, R11, J10, K5, D5, C2, R12, change R13 for a 10R resistor and link point 4 to point 3 on the K5 slot?

Welcome Niko.
Yes (R13 for 10K and not populate R12), or keep both 20Ks R13 and R12 and link pin 9 with 10 at the K5 footprint as well.

 

[Harpo]

Added a 5K resistor in series with VR17 and all is well in the land of gain reduction metering. For those who want or need this fix, here is what I did. I cut the trace on the back side of the PCB between C121 and VR17 Pin 1 (CCW) and then soldered the resistor on the backside between those points. This allowed me to correctly complete Calibration Step 4) Meter/Ratios with the 1ma (slightly less in reality) Hairball compression meter.

"for an easier fix an about 3K3...4K7 fixed resistor at the meter terminal" would have done the same without cutting a pcb trace.