Bus Testing

Using the channel testing technique described in the previous post, I soon identified a problem with the first Helios channel I tested. It was OK when the EQ was bypassed but when it was switched in, the level dropped by about 30dB. Clearly something was not right. A quick inspection of the PCB near where the signal is input to the EQ revealed a simple wiring error. I corrected this and the EQ worked as expected.

So, with the channel testing set up and working and with a working channel to play with, it was time to start testing the buses. To do this, all that was necessary was to wire up the buses to carry the mixed signal to the mix amplifiers, and to connect the Left and Right mix bus to the mix amplifiers via the master volume control. (There was no need to do this for the AUX sends as their master controls are already wired up in the AUX send module). The picture above shows the new test set up. The input, output and channel fader are as before. The added bus wiring and the master (stereo) level control are indicated on the picture above.

Before adding the stereo master level, I checked the AUX send buses. It was a good job I did because I soon discovered I had wired the AUX send module to the L&R master buses instead of the AUX send buses. Changing the connections of two wires in the AUX send module fixed this. I then connected the master level control and tested the stereo bus. The Smart Pan worked exactly as expected and I measured the Pan mid drop as 4.3dB, very close to the calculated value. In fact it all went very smoothly.

As part  of building the test set up, I had to do quite a lot of crimping and I have to say that, with practice, it becomes quite easy. You need to use the right sort of wire and I found that lightly tinning stranded wire was a great help but other than that it is becoming quite straightforward. I might at some point in the future revise the motherboard slightly to change the bus connectors from Molex to IDC for two reasons. First, it is easier to make a cable of an accurate length using IDC and secondly, it would allow me to include a ground wire between each pair of bus wires, thus reducing bus to bus capacitance and hence reducing high frequency cross talk. As yet I have no reason to suspect there is a high frequency cross talk problem but so far I have only been making tests at 1KHz.

Channel Testing

Now we have some channels fully built, it is time to test and calibrate them. The picture above shows the set up used to do this. It shows the main sub-rack from the rear. You can see the three motherboards into which the modules plug, each motherboard taking two modules. Towards the bottom of each motherboard you can see the power bus wiring that links the heater and HT voltages between motherboards which are made using heavy duty terminal blocks. All the signal connections to and from each module are made to Molex connectors soldered to the rear of the motherboards. Connections from these to other parts of the mixer are made using crimp connection Molex plugs (see earlier entry on crimping).

For module testing we first need an input connection. This comes from the XLR connector labelled 'Input' which connects to a 3 way Molex crimp connector that can be plugged into the mic or line input Molex on the motherboard. The output comes from another Molex crimp connector to a VTB2291 output transformer and then to the XLR labelled 'Output'. These two are all you need to test the first amplifier stage in a channel. You plug the input into either the mic or line input and take the output from OUT1 of the module.

In normal operation the OUT1 signal goes to the channel fader and the return from the fader is connected to the Fader input on the motherboard. This input is then connected internally to the input of the EQ section of the channel and the EQ output goes to the input of the second amplifier (IN2) and the channel output comes from OUT2. So for testing the whole module you need the additional connections shown to and from a fader and the VTB2291 is connected to OUT2 instead of OUT1. Thanks to Tony Reeves of  MTR for the bargain 100mm throw faders at just £2 each

I started testing the Helios channel module and the first stage worked fine. However, with the fader connected, the signal drops a lot when the EQ is switched in. Clearly there is a fault on the EQ board that needs to be fixed.

Pultec 3 Band EQ Channel Completed

I have now finished building the Pultec channel module. The overall topology is the same as the Helios 69 channel amplifier described earlier, the only difference is in the passive EQ. The Pultec EQ is based on the EQP-1A with an added mid boost/cut. This EQ has a lot more controls than the Helios 69 and takes rather longer to wire up. The EQ in/out switch is more complex to wire up as well as it uses a separate pad. Building this has convinced me more than ever that reducing the amount of hand wiring would be a good objective. It's not much fun, it's hard to keep it tidy and it is prone to errors.

Now this module is complete I can mate it and the Helios 69 channel with the bus amplifiers and check out the basic functionality of the mixer.

Helios 69 Channel Amplifier Completed

The picture above shows a completed channel amplifier incorporating a Helios 69 style EQ. This is instructive for illustrating the amount of hand wiring required to complete a typical module. Some of this is unavoidable and some of it could be simplified as will become apparent.

The PCB on the right is the channel amplifier which contains two complete valve amplifiers. The PCB on the left is the passive Helios Type 69 EQ. All connections to and from the channel are via the 32 way DIN connector at the bottom right of the picture. I will describe the overall signal path and the wiring associated with each part of it.

Mic and Line signals enter through the 32 way DIN connector and connect to nearby pads. They need to connect to the push buttons on the front panel. Rather than track these across the PCB and risk picking up interference in these low level signals, they are run across the board in the screened cables at the right of the channel amplifier. From the push buttons the mic and line signals are routed via the microphone transfomer to the first valve amplifier the output of which goes to the 32 way DIN. This signal normally would feed an external fader. In this mixer the intention is for it to feed an unbalanced insert, the return from which feeds the fader. The return signal from the fader is connected back to two pins on the 32 way DIN (labelled Fader return in the picture above). From here they are wired via a screened cable to the input of the EQ PCB.

Three of the Helios EQ controls are on the PCB itself. The remainder, the bass boost pot, the peak/trough toggle switch, the mid gain pot and the EQ in/out toggle switch all need to be hand wired to the PCB. This is not particularly difficult or awkward but this is a relatively simple EQ. The Pultec EQ, for example, has six pots and two switches to wire up i.e. twice as many as the Helios EQ. Both EQs would be easier to wire up if the off board controls were all on their own PCB with just a single plug to connect to the EQ board itself.

The output of the EQ is connected to the second valve amplifier which has a preset gain sufficient to make up for the loss in the passive EQ and to allow for 10dB in hand on the channel fader. The output of this amplifier appears on the 32 way DIN and also at a pad labelled Out 2 above. The DIN connector output is used if a direct output is required or if an external routing module is used. In this case the routing is contained within the channel module so a screened connection is made from OUT2 to the pan pot (see picture). This channel module uses the Smart Pan which allows two toggle switches to route to individual channels or engage the pan pot or to mute the channel. The wiring for this would be quite complex so a small PCB was designed to do this. It is split into two halves - one for the pan pot and the other for the two routing toggle switches. Just four wires are required to connect the pan pot PCB to the Smart Pan toggle switch PCB. Two wires, one for left and one for right, are connected from the Smart Pan PCB to the bus resistors which then connect to the 32 way DIN.

Lastly, 0V and OUT2 signals from the Smart Pan PCB are hand wired to the two AUX send pots (both of these are post fader sends). Individual wires from the sliders of these pots are wired to the remaining two bus resistors.

Improvements

Construction would be simplified if fewer hand wired connections needed to be made. Some, like the mic and line screened cables, are unavoidable but could perhaps be simplified by using connectors. The EQ would certainly benefit from a PCB to hold all the controls that will not fit on the main EQ PCB and this should be connected to the EQ PCB via a single cable - possibly a ribbon would do.

A small connector for Out 2 would be useful as would combining all the pan, routing and AUX send components onto a single PCB. This would need a 2 pin connector to Out 2 (via a screened wire) and a 4 pin connector to the mix bus resistors (possibly a ribbon).

At the moment it is not possible to set the gain of the second valve amplifier with the modules plugged in so another improvement would be to move that component near to the front panel so it could be accessed through the front panel with a screwdriver (thanks to Holger for this suggestion).

Grayhill 71 Series Switch Stops

Grayhill 71 Series Switch Stops

The demo mixer uses a lot of PCB mounted 71 series Grayhill switches. These are all basically 12 position switches but they can be adjusted for fewer positions if required. This is done by inserting one or more pins into small holes in the front of the switch to act as end stops. To stop them falling out, Grayhill provides a self-adhesive cover. With no pins fitted the switch just goes round and round which can be useful in some circumstances.

Fortunately, every time I have used the Grayhill 71 series switches, I always laid out the PCBs starting from position 1 so that if any positions were left over they always started at 12 and worked downwards. This means there is one standard way of adding the stops to the Grayhills that only varies with the number of positions required.

Even if all 12 positions are used, you still need to add one pin if you want to stop the switch continuing from position 12 back to position 1. This is exactly the case with the mic pre gain control where the maximum gain position is 12 and the minimum gain position is 1. The last thing you want is to accidentally switch from minimum to maximum gain. So for 12 positions you need a single pin and for fewer positions you need two of them. The diagram below shows how to locate the right holes into which to fit the pins and lists the various switches used in the mixer and their corresponding pin positions.

I Hate Mechanics

I willingly admit I am not good at mechanics. I can't saw a straight line or reliably drill a hole on the right place, but sometimes mechanics is unavoidable. Below the modules of the mixer is a 2U panel where the channel rotary faders are to be. It really needs to be made of aluminium so it can be engraved but even 3mm thick aluminium is not strong enough such that it it can reliably span 19 inches and survive being leant on - I know I accidentally did it.

A long time ago, when I was building another mixer that was intended to use the same RackZ 19 inch console housing as this one, I had some special wide aluminium extrusions made that were designed to fit right across the RackZ console. I never used them and still have them so I thought I might be able to use them to add strength to the aluminium panel. Unfortunately, when I 'designed' the extrusion, I forgot to account for the 19 inch rack panel mounting built in to the RackZ console so they don't fit!

However, I have a lot of other bits of extrusion accumulated over the years so I had a rummage through them and found a couple that seemed ideal. Each is just shorter than the panel width and is about 6mm thick and 12mm wide. More importantly, they have several 3mm tapped holes along their length, ideal for attaching to a panel. So I stuck them to the panel with Sellotape and used a small drill to make pilot holes in the front panel where the tapped holes were. I removed the extrusions, drilled out the holes in the panel to 4mm and then bolted the extrusion to the panel. I am pleased to say the result is a very strong assembly that, when fitted to the RackZ console, can be leant on without fear of it bending.

I don't like doing mechanics but occasionally you win.

Mix Amps

The mix amps are nearing completion. These amps are responsible for raising the mix bus signals back up to line level. The Twin Line Amp (TLA) was designed specifically with this in mind and two such boards are used to provide the four required bus amps - two for the L&R master and two more for the two AUX buses. The PCBs are shown below:

The TLA board is very similar to the mic preamps. The same three tubes are used in the same topology. The main differences are that in the TLA the mic/line/pad/phase/48V switches have been removed and the stepped gain control has been replaced by a trimmer. The space made available by these changes has allowed an extra input transformer to be fitted. The TLA therefore provides two uncommitted line amplifiers with gain variable from 6dB to 40dB and a pair of uncommitted input transformers. The transformer footprint allows 10K:10K bridging or 1:10 mic input transformers by Cinemag, Jensen or Sowter to be fitted.

For operation as a mix amp no transformer is needed. The AUX send mix amp is the PCB at the bottom of the picture. The bus inputs from the edge connector are simply connected direct to the AUX send master pots (each 10K log) and thence to an amplifier which provides the necessary gain make up to compensate for the passive mix bus loss and any gain 'in hand' required for the master fader plus any loss in the output transformer. With four channels feeding the AUX buses via 47K resistors and with a 10K master AUX pot, the total bus loss is close to 19dB. So with 0dB in hand on the AUX send master control, the mix amp gain would need to be set to 19dB. However, since a balanced output is required, a 2K4:600 transformer is fitted externally which drops the level by 6dB. A total gain of 19 +6 = 25dB is therefore required.

The upper board is the main L&R bus mix amp. On the front panel of this board are the AUX return controls consisting of level and pan controls for each of the two AUX returns. The reason they are on the main L&R bus mix amp is because the AUX returns need to be fed to the L&R mix buses so this is a convenient place to put them. Again a TLA board is used but this time a pair of 10K:10K input transformers is fitted to provide balanced input to the AUX returns. The transformer outputs feed the level and pan controls, the output of which feeds the L&R mix buses via 47K resistors. The two amplifiers on this board act as the master L&R mix amps. As before the mix bus signals end up at the master level control but in this case the master L&R fader is external to the module. As before there are four channels feeding the master L&R buses but in this case the AUX returns also feed it. With 47K bus feed resistors and a 10K master fader, the the total bus loss is just under 21dB and if we need a balanced output then the make up gain will need to be set to about 21 + 6 = 27dB

The AUX send board is complete and ready for testing. The master L&R board is waiting for the pan pots to arrive.