Allen & Heath 621 Vintage 6-channel Mixer
Recently acquired and in need of some love:
Allen & Heath product page
The 621 is the baby of the 21 Series stereo mixers that AHB released
about the same time as the larger System 8. It shares many common features
both mechanically and electronically -- the input stage is pretty much the same,
and it sports the same brown-and-wood theme as well the same knobs, button
caps and VU meters of the MkI and MkII System 8s. They even share the same MPS8 power supply.
A convenient benefit of pairing this 621 with a System 8 is that it is
very easy to link the two mixers. With just three stereo ¼" jack leads
the 621's stereo bus, two auxes (CUE is pre-fade, ECHO is post-fade),
and PFL audio and control, can be connected to the System 8's
The one major issue with these vintage mixers is that, similar to the MkI
System 8 (and later changed in MkII and MkIII), they use 90mm TBM 5kΩ faders.
Rare as hens teeth, so for worn faders the options are to rebuild
or try to fit standard 100mm faders.
The 21 Series was clearly designed to meet a lower price point than the System 8.
On the System 8 we have an XLR for the mic input and a ¼" jack for the line input.
The 621 provides only a single XLR for both mic and line.
However the way that AHB chose to wire their insert jacks (ring=send, tip=return)
makes it possible to use the inserts as
line-level (+4dBu) inputs, bypassing the pre-amp.
There are various changes to be investigated, including liberal use of
NE5532 op-amps, capacitor refreshing, and so on, as well as general tidying-up.
The general plan is to use this smaller mixer to determine the more common issues
of this type of mixer, and experiment with some improvements, prior to starting
work on it's bigger brother.
None of the ICs have local decoupling. Other than the 100uF on each rail at
the point of entry they must be relying on the PCB traces having such low
inductance as not to be a problem. Hmmmm.....
So far it looks like all the electrolytics were manufactured in 1983.
Which makes them 30 years old this year! Time for a celebratory cake.
And time to replace them with some fresh new Panasonic FM-series capacitors.
The 10uF 16V capacitors in the audio path will be upgraded to 47uF 25V
(Allen and Heath had already made this change themselves by time of the GS3 in 1991).
The exceptions are the timing cap in the peak detectors on the input channels,
the VU meter timing capacitors,
the much larger 220uF ones in the pre-amps,
and the two phantom-power blocking
capacitors -- these will be replaced with new capacitors of the same value.
- Replace IC1, IC4, IC6 with NE5532 - DC blocking caps already in place, but check polarity.
- Change power LED to bi color to indicate when phantom power is on (Rapid 56-0346)
with green for ON and orange (green+red) for ON+PHANTOM.
Note because the in-mixer power switch is only two-pole we won't
be able to isolate the phantom power with it. So put a switch in the PSU (see below)
so it can be turned off. Will need to reconfigure the normal power LED
to allow LED to work when power disconnected - reverse schottky diode.
Phantom power will then show as red on the power LED.
- Add ground wire to anti-shake washer to cut through paint to make good
contact with metal panel and earth it. Also check earth bonding between
top panel, baseplate and side plates, add wire jumpers if necessary.
- Try out experimental VU meter lamp soft-start dimmer.
- Try out balanced line driver (either THAT1646 or NE5532).
- Could the unused monitor option (when all switches are out) be used?
Is there some other internal signal that would be worth monitoring?
Or is "off" a good thing to keep?
- Replace IC1, IC2 with NE5532 -- fit 10/16 in series with R16,
check polarity of other DC blocking caps.
- Try out balanced line driver.
- Replace IC1, IC2 with OPA2134.
- Replace Q1-4 (ZTX214, NF=10dB) with 2N5087 (NF=2dB).
- Fit phantom power resistors to channel 5 and 6.
- Fit bleed resistors just before the insert send.
The Curious Case of C15
The channels use a 5kΩ fader to feed both the pan and the post-fade
aux ("Echo"). The PFL taps off at the fader input, as does the peak
detector and optionally also the pre-fade aux ("Cue"). So the worst-case
load on the last op-amp IC2A is actually quite low, which has important consequences
for one little capacitor - C15 - both in terms of -3dB point and also distortion.
As specified, that little 10uF capacitor gives a worst-case -3dB point at 10.7Hz.
Not hugely bad, but could be better. More importantly, going on the work that
Doug Self has done in examining distortion in electrolytic capacitors we must
aim to keep the voltage across electrolytic capacitors below 80mV. In the case
of C15 at worst load with optimal +4dBu signal that point is at 164Hz.
The quick and obvious choice is to change C15 to 100uF 25V as already used for
supply decoupling duties. This would bring the -3dB point down to about 1Hz,
and bring the frequency at which distortion may become noticeable down to around
16Hz. Much better.
Note: for those thinking that it is silly to aim for a -3dB point at
anything below 20Hz in an
audio circuit, you need to add up the contributions of every high-pass filter
in the complete input-to-output chain. In the input channel alone you have
something like five DC blocking capacitors (undesirable high-pass filters),
followed by a further three in the left/right outputs, and another four in front
of the mono output. That's a maximum of twelve little filters eating away
at the bottom end and contributing phase errors.
An interesting observation over the weekend: there is a space on the input
channel PCB for a rail decoupling capacitor close to the op-amps. Was this put
there as an insurance policy, and not fitted in production?
Close inspection of the PCBs showed some changes between them
and what is shown in the schematics:
- Monitor source switches are in a different order: if you look at the schematic
the position of the tape and mono switches is swapped, giving the tape
input highest priority - which is actually what the user manual states!
- There is a place on the channel input PCB for a mute switch, but no switch is fitted
and no cutout on the panel, so a link is fitted in it's place. Could
this have been for a Mk.II "feature" if this line of desks had proved
successful? It would have only needed a change to the front panel and one
extra part per input channel. I think retro-fitting a mute switch would be
too invasive on the existing front panel.
- Also on the channel input PCB are pads for a separate
line-in jack socket. A build option chooses between this or sharing
the XLR. Again, was this for a Mk.II upgrade "feature"?
If it did, would they increase the depth of the back panel to make room
for another jack socket? Otherwise it looks a bit tight.
The first phase is to break the mixer down into its component parts, ready
for cleaning prior to any repairs and/or modifications.
Photo Album of the 621 Decontruction
After playing around with an input channel, just for fun of course, the first PCB to get
up and running is the master module, since this has the headphone amp and VU meter drivers
so I can start to see and hear what is going on.
I'm recapping with minty-fresh Panasonc FMs, upgrading some TL072s with NE5532s, and
generally cleaning it up. Here it is just before work starts:
A quick test setup later and the headphone amp is found working, as is the PFL
relay (invariably a sticking point), the tape input pre-amp, and the monitor selector
switches. I think I'll replace the two level pots with new ones while I have the board
on the bench and the local Maplin store has some in stock (FX09K for those watching at home).
Here you can see the new Panasonic FM electrolytic caps, the NE5532 upgrade to
the stereo monitor output driver, and the headphone level pot - you can just about make
out the OMEG brand name.
The 21 Series mixers, together with the System 8 Mk I, used 90mm 5k log taper
faders from TBM. While TBM still exist, they no longer make the faders, and
replacements are impossible to find. What to do, eh?
Simples! Make up an adaptor plate that allows a standard 100mm fader
(I used Bourns) to fit into the same space. Here you can see the parts ready
Now, the smart-asses amongst you will point out that 5k log faders are not common.
Indeed not. So that's why I use a stereo 10k log fader and connect the two halves in
parallel, which behaves exactly the same as a 5k log fader.
More advanced smart-asses will then point out that using a 100mm fader in a 90mm
slot means we lose the last 10% of wiper travel. And indeed we do. But on a log
fader 90% of travel translates to about 95% of the resistance curve. What that means
in practice is an error of -0.5dB, so the "+10" marking now means "+9.5".
Only the truly anal will complain about that. The rest of us will enjoy shiny new
faders in our vintage mixers.
At long last I have finished the internal work! The PCBs are back, the
board links in place, and everything seems to work ok.
Of particular note is the small panel of Veroboard between the output boards
on the left and the channel inputs on the right which makes it easier to separate
the two sections in future, and also neatens up the wired connections from the back panel.
On the input channels you can see where I have replaced all the electrolytics
with Pansonic FM types:
First up is giving the knobs a thorough clean in warm soapy water. Most of the
caps were removed - the few that didn't seemed to be glued in with something like
superglue (cyanoacrylate)... annoying. Some of the others were held in place with
blutak, but they cleaned up nicely.
Interesting to note that the knob colours on arrival (see above) do not match
the AHB brochure. Apart from some missing caps, the biggest difference is in the
CUE and ECHO master pots, which both had white caps, while they should have brown
caps. While I do have some spares in stock for the big mixer restoration project
I think I will keep the white caps for now and see how it goes.
My original plan was to design a new power supply for this mixer. However
when you factor in how much time it would take, I decided to buy a PowerOne
linear power module and put it in a box I had in my parts store. Much quicker!
Finally, after all the work has been done, the mixer is finally finished: