ESP Project 30

[ramshackles]

I was wondering if anyone has ever built any of the modules or indeed a mixer from "project 30" on the ESP pages:
http://sound.westhost.com/project30a.htm

I was thinking about doing a simple N into 2 mixer, without the tone control section and I have a few questions. Attached is a redrawn schematic (copied straight from ESP) of the transformer balanced input stage and the fader/pan stage.

Here are my questions

Input stage:
1. I noticed that the transformer driving a 5534 opamp for gain is quite similar to the ISA110 input stage, although the 110 uses 2 5534's; one for a stepped gain and a further stage for trimming. Would there be any noticeable advantage to adjusting the circuit to something more similar to the 110?

2. Are there any alternative transformers to the pricey JT-16-A? I've had a look on lundahl & cinemag and the only possibilitie I've come up with would be this:
http://cinemag.biz/mic_input/PDF/CMMI-2C.pdf

Fader stage:
1. I'm guessing it would be fine to just tap off as many auxes as desired in the same manner given in the original schematic
2. The channel insert point could just become a direct output option (as there is no eq) with no changes in the schematic - am I right here?
3. In small signal audio, Douglas Self is pretty scathing about the type of panpot setup used here. I include his proffered panpot solution as a second option - it seems like there would be no problem in just 'plugging' in straight in, provided the routing/mix resistors are then set correctly

Anyone had a look at this project?

 

[ramshackles]

I decided to try and build a 'simple' preamp using the modules described in project 30.

A few minor changes I made are using Lundahl transformers (LL1527 for input ad LL1517 for output) and replacing the 50K gain pot with a stepped switch (Lorlin) giving a total of 100K.

I'm not sure I have drawn up the switch connections correctly, however.

Next I'll start working on a layout for the thing

 

[Harpo]

A few minor changes I made are using Lundahl transformers (LL1527 for input ad LL1517 for output) and replacing the 50K gain pot with a stepped switch (Lorlin) giving a total of 100K.
I'm not sure I have drawn up the switch connections correctly, however.

Starting from left side of schematic ...
SW2 shows pad engaged for its default setting. For usual it is the other way round.
SW3 shows polarity reversed for its default setting. For usual it is the other way round.

The LL1527 primary and secondary connections look a little mixed up.
You probably want the primary windings in parallel (but not self canceling, have a look at the winding dots) so join pins 2+4 and pins 1+3.
You probably want the secondary windings in series (but not self canceling, again look at the winding dots) so join pins 6+7, pin 8 to opamps non-inverting input, pin 5 to reference voltage. For your layout you might fit a LL1527XL as well, but watch out for transformers larger outer dimensions.

R16+C7 might only be a placeholder for a zobel network when using a different transformer. For usual no zobel required for the LL1527.

Your resistors series string at the switch doesn't add up to 100K but to 87K7 and the stepwidth is -maybe on purpose- not a linear dB increase (from +0.83dB for min.gain setting you add +1.93,+3.16,+3.56,+3.69,+3.49,+3.31,+3.88,+4.30,+4.58,+6.20dB delta between following steps) for a total of +39dB at the opamp gain stage, add +6dB from the input transformer on top. (I'd scale the Rfb/R23 resistor network down for lower resistive noise at cost of a larger C8 and vary the shunt arm resistor value instead of the feedback resistor value, but YMMV.)
The connection between throw-pin1 and pole-pin13 is missing, giving a maybe no feedback resistor condition between steps while switching, causing the opamp to run open loop only limited by the voltage rails.
A placeholder for a (maybe 22pF for 100K Rfb) compensation cap across this feedback resistor might be a nice to have, if so required for highest gain setting. With decreasing Rfb this cap will be pointless.

IMHO IC2 seems pointless. I'd place R49 before C9 and omit R48, IC2, C5, C10.

Lots of parts and trace connections are missing a junction dot in your schematic.

From your previous post 'I'm guessing it would be fine to just tap off as many auxes as desired in the same manner given in the original schematic'
you might guess wrong when you see the load resistance of all these paralleled resistors (and all bus resistors probably connect to a virtual ground node as well) and the current drive ability limits of a NE5534.

just my 2ct and good luck.

 

[ramshackles]

Thanks for the reply. I just followed the datasheet numbering for wiring the 1527 in 1:2 configuration, which was 3+2 and 4+1 on the primary and 7+5 on the secondary with pin 8 to +ve and pin 6 to. But looking closely, it looks like it should be 1, 4 and 5, 8 that are +ve, so the symbol I've found for it isn't accurate. I've changed the dots around.

Yes, I wasn't sure about still including the zobel network, but I think I'll leave it in incase of a different transformer.

Yes the resistor series is purposefully not linear - I don't know the normal way to do it so I just went for what I thought would be most useful with the closest E96 values. I'v fixed the missing resistor and it should now add up to 100K

So you would vary R23 and keep the current switch fixed at 100K? What would be the benefit? Scaling down it could be 10000K fb and 100 R23, but would that need something like 470uF for the capacitor?

I think the original idea behind IC2 and the output stage was that there would be perhaps an EQ stage or even further stages (fader etc) between the preamp and the output...although I'm not sure this would warrant IC2 at all.

The original idea of ESP 30 was to have a tone stage, where I have the take out point for a peak indicator, and this going on to a fader, pan and routing stage.

For aux buses (as in the original esp mixer idea), if the mix resistors were suitable scaled up would this not suffice? I think the original has 2 aux buses with 22K mix resistors.

If one were to have 6 aux buses, but make each pair switchable (so that each channel could only access a max of 3 aux buses), you could then use something like 47K mix resistors?

I've attached V2 of the preamp schematic. IC2 not removed just yet....

 

[Harpo]

I just followed the datasheet numbering for wiring the 1527 in 1:2 configuration, which was 3+2 and 4+1 on the primary and 7+5 on the secondary with pin 8 to +ve and pin 6 to. But looking closely, it looks like it should be 1, 4 and 5, 8 that are +ve, so the symbol I've found for it isn't accurate. I've changed the dots around.

The symbol is for a LL1527A, whatever this pinout might be. Your schematic and wording is a LL1527 or LL1527XL. Follow the pinout of the transformer you intend to use and be aware pin spacing and/or parts outer dimensions might differ from your Kicad library part.

Yes the resistor series is purposefully not linear - I don't know the normal way to do it so I just went for what I thought would be most useful with the closest E96 values. I'v fixed the missing resistor and it should now add up to 100K

So you would vary R23 and keep the current switch fixed at 100K? What would be the benefit? Scaling down it could be 10000K fb and 100 R23, but would that need something like 470uF for the capacitor?

You now changed the schematic for a different Rfb (now 1K instead of previously wanted 100K) and different gain setting by varying the shunt arm resistor between zero ohm (bang) at pin12 and the IMHO too high sum of your resistor values at the steps. DC voltage gain is set by 1+Rfb/Rshunt or for a dB readout LOG(1+Rfb/Rshunt)*20. The 1K Rfb might be a little optimistic, usual value for Rfb would be 10K ... 20K. Picking 20K for Rfb and the wanted gain range up to +40dB plus transformer gain, min.Rshunt would be 200R for +40dB and the resistive string would add up to maybe the value of Rfb for +6dB. For AC voltage gain you have C8, forming a 1st.order (-6dB/oct. -20dB/dec.) HPF with -3dB cutoff frequency set at 1/(2*PI()*Rshunt*C8) with Rshunt in ohms and C8 in Farad. For an assumed wanted linear audio frequency range of interest between 20Hz to 20kHz you might scale the value of C8 for a decade below 20Hz, IE 2Hz. With highest gain setting (lowest value of Rshunt), C8 would be 397uF for 2Hz and Rshunt 200R. A common value 470uF will set cutoff at 1.7Hz for a phase response of -4.8° tested at 20Hz.

I think the original idea behind IC2 and the output stage was that there would be perhaps an EQ stage or even further stages (fader etc) between the preamp and the output...although I'm not sure this would warrant IC2 at all.

The original idea has an insert between these stages, and not knowing the current drive ability of whatever might get plugged in, IC2 is operating as a unity gain buffer to drive the panning circuit and the Auxes.

The original idea of ESP 30 was to have a tone stage, where I have the take out point for a peak indicator, and this going on to a fader, pan and routing stage.

Your description is missing the buffering +6dB gain stage after the fader.

For aux buses (as in the original esp mixer idea), if the mix resistors were suitable scaled up would this not suffice? I think the original has 2 aux buses with 22K mix resistors.

If one were to have 6 aux buses, but make each pair switchable (so that each channel could only access a max of 3 aux buses), you could then use something like 47K mix resistors?

R12+R13 load opamp by 20K. Panned to an extreme setting, either R5 or R6 connect to 0V reference voltage, giving 3K3 load, pan other side has 3K3 in series to the paralleled 10K panpot and 22K bus resistor. Total load of ESPs pan circuit is 3K3||(3K3+10K||22K), giving 2K491.
Load of Aux1 and Aux2 with pot in full CW position and Sw1 in post fader setting is 10K||22K||10K||22K, giving 3K437. These Auxes are in parallel to the pan load, so U1 has to drive 1K347 and will have no problem to do so.
From your 1st.schematic you had 4 additional Auxes, giving a total resistive load of 755 ohm (ignoring the load of your not shown VU circuit). At some point you will exceed limits of opamps current drive ability. You could increase resistor values at cost of increased resistive noise or add buffers per bus at cost of cash and size. Pick your poison.

 

[ramshackles]

So, new version.
Rshunt is now varied from a min of 200R, with Rfb fixed at 20K. IC2 removed, along with the 10u cap before the peak detector and R48, C5, C10.

Am I right in thinking the switch is still wired wrong and that R4 should be connected to pin 12 directly, R7 to pin 11 and so on? (i.e. each resistor shifted along to the next position, leaving pin 13 connected directly to pin 1?)

Is the purpose of having the cutoff frequency (1.7Hz for 470uF C8) some way below the freq. of interest to avoid much phase changes?

I suppose that, if there were to be additional stages (taken from the same place as the peak detector) such as fader, aux or whatever, you would put the 10u cap before the peak detector back?

 

[Harpo]

Rshunt is now varied from a min of 200R

From your schematic from zero ohm at pin12, up to the sum of your resistive string adding up to 29K182 instead of 38K492 because of shorted out R24. You probably want to revisit these resistor values anyway.

..along with the 10u cap before the peak detector and R48, C5, C10.

My reply was for schematic Line_Mic_Module[1].pdf. You deleted C9 and renamed C10 to C9. With Lundahl LL1517 static resistance of 9.2 ohms per winding + 100R from R49, value of C9 better be at least 768uF, so probably a 1000uF instead of 10uF, else you set your 3rd. HPF @135Hz.

Am I right in thinking the switch is still wired wrong and that R4 should be connected to pin 12 directly, R7 to pin 11 and so on? (i.e. each resistor shifted along to the next position, leaving pin 13 connected directly to pin 1?)

Yupp. I'd suggest you fill in your parts values to calculate the amount of expected gain from previous post already given formula [LOG(1+Rfb/Rshunt)*20]. For example LOG(1+20000/(200+365))*20=31.22dB for step position 11.

Is the purpose of having the cutoff frequency (1.7Hz for 470uF C8) some way below the freq. of interest to avoid much phase changes?

At cutoff frequency [freq=1/(2*PI()*R*C)] signal will be decreased by -3dB (voltage gain of 0.707 or 1/SQRT(2)). For a HPF, this is a phase response angle of [-ARCTAN(cutoff_frequency/test_frequency)*180/PI()], giving -45° when cutoff frequency and tested frequency are the same. Assuming your frequency range of interest above 20Hz, you want this least affected by these filters, so you try to keep phase response angle to not exceed -5° by decreasing the cutoff frequency. The critical part is the lowest resistance (200R from your schematic) for the highest gain setting. You might argue, only a mic might use this amount of gain and you maybe don't want a mic to boost this rumble frequency, so a lower value cap might be sufficient. Using this preamp as a makeup gain stage for a passive summer might manifest the need of this cap value instead.
From your schematic you already have 3 highpass filters implemented, the 1st. set @6.8Hz by C2 and C3 with reflected impedance of R15. Getting rid of the IMHO not needed 1st.HPF with phantom blocking T1 fitted, I'd omit these caps and the (parts values missing) zeners in front of T1. YMMV.

I suppose that, if there were to be additional stages (taken from the same place as the peak detector) such as fader, aux or whatever, you would put the 10u cap before the peak detector back?

From my reply to schematic Line_Mic_Module[1].pdf, the DC blocking cap was before the peak detector and R33 after the cap.

 

[ramshackles]

Ok, I think I understand my mistakes.

Forgetting the values of the switched resistors, I've just tried to put them in the right place this time.

R49, R33..now unsure of their values or necessity. R33 would form a HPF with C9?


On ESP 30, it is mentioned that the zener diodes can be omitted, as you stated (so I have done this), but it also says they are recommended anyway; for what purpose?

 

[ramshackles]

Gain switch drawn more clearly...

 

[Harpo]

Forgetting the values of the switched resistors, I've just tried to put them in the right place this time.

dB values are upside down, IE 40dB at the lowest shunt arm resistor setting and vice versa.
By C1 22pF between NE5534-pins5/8 (not one side connected) you already slowed down this opamp for unity gain stable operation, else a MBB switch or a higher value resistor across this switch would be needed to avoid an open connection between gain settings. I wouldn't call LPF setting C2 optional.

R49, R33..now unsure of their values or necessity. R33 would form a HPF with C9?

R49 limits current load and is part of a HPF. R33 is a (negliable) part of HPF with dominant T2 static resistance in parallel fitted. Optionally omitting T2, R33 will also be bias resistor for your peak showing circuit. Could be left out with T2 fitted.

On ESP 30, it is mentioned that the zener diodes can be omitted, as you stated (so I have done this), but it also says they are recommended anyway; for what purpose?

Relates to the phantom blocking 47uF caps. (Your T1 blocks phantom DCV). When these caps are charged from +48V phantom, they limit voltage connected to opamps input terminals below opamps supply voltage when phantom voltage is switched off and these caps dump back charged energy.

 

[ramshackles]

dB values are upside down, IE 40dB at the lowest shunt arm resistor setting and vice versa.
By C1 22pF between NE5534-pins5/8 (not one side connected) you already slowed down this opamp for unity gain stable operation, else a MBB switch or a higher value resistor across this switch would be needed to avoid an open connection between gain settings. I wouldn't call LPF setting C2 optional.

Oops I think it was an error in exporting to PDF that it isn't connected to pin 8.
so...C2 at around 100pf would give a LPF at ~80KHz?

R49 limits current load and is part of a HPF. R33 is a (negliable) part of HPF with dominant T2 static resistance in parallel fitted. Optionally omitting T2, R33 will also be bias resistor for your peak showing circuit. Could be left out with T2 fitted.

Perhaps a place holder for R33 for if T2 is omitted.

 

[Harpo]

C2 at around 100pf would give a LPF at ~80KHz?

Yupp. 79.577kHz. NE5534 has a GBW of typ.10MHz with 22pF compensation cap fitted. Divide this by your max.gain setting of 101, you better stay below 99kHz to keep the opamp in its safe operating area.

 

[ramshackles]

I threw in a possible HPF. (note the rotary switch on this schematic is marked wrong. Should be a 6 position, single deck)

I started drawing up a PCB layout. I wanted it on a eurocard, although this leaves a lot of free space. An instrument input could be an option.
I put the ground connections as tracks to be sure that they will all link up, ended up needing 2 short segments to be in the other layer. I don't know if there is any real need for a ground plane...

All power tracks are also on the ground layer and audio tracks have largely been kept to the top layer (a few short tracks for switches on the btm layer).