Yes, that would be OK. The input impedance here is 500K so for a 75Hz -3dB point the capacitor should be 4.2nF. There nearest standard value should be OK.
VR3 should be on the PCB. It should be a pre-set type and adjusted during test.
Either of these will do. The first has a 50K input impedance but drives an unbuffered half wave rectifier diode. The second has a 10K input impedance and a buffered full wave rectifier. On balance I prefer the second circuit but I would modify it to increase its input impedance to 100K by changing R1 and R2 to 100K.
The TL082 normally needs both positive and negative supply connections. If you have a -12V supply for it then this circuit as it stands is OK and needs no input capacitor. If you wish to operate it from a single +12V supply then you need to bias the + input to about half the supply voltage. You can do this, for example, by adding a 1Meg resistor from the + input to the +12V but you will then need an input capacitor.
I find wiring panels to PCBs to be one of the most awkward and tedious tasks I can imagine. Screened leads are obviously the way to go but I would recommend reducing the wiring as much as possible. PCBs holding pots and switches that mount directly onto the panel can save a lot of time and effort and improve reliability too.
Cheers
Ian
1)
Why 500K? ... there is a R of 1M to ground.
2)
- Could the input impedance of 100K (with R1 and R2 = 100k as you suggest) influence the main circuit (it is in parallel with the 50K fader)?
- Could we utilize a TL082 in substitution of TL071, in order to employ a single component for both vu-meter and buffer for monitors or it is preferible two independent components ... TL071 and TL081 ? In the first case, the vu-meter circuit needs some modifications?
Ian,
Why 500K? ... there is a R of 1M to ground.
2)
- Can the input C1 (10uF) of the vu-meter be eliminated?
- Could the input impedance of 100K (with R1 and R2 = 100k as you suggest) influence the main circuit (it is in parallel with the 50K fader)?
- Could we utilize a TL082 in substitution of TL071, in order to employ a single component for both vu-meter and buffer for monitors or it is preferible two independent components ... TL071 and TL081 ? In the first case, the vu-meter circuit needs some modifications?
Last edited:
You forgot transport. Yes, of course you're right for the speakers, although they're quite a bit cheaper (relatively) than in my youth. But the work of the sound technician does not need to be anymore complicated at all, just a trifle more setup time. If you get the layout of your control panel right, it's not immediately obvious you're running on a non-standard system (think post fade insert points on every channel feeding individual stacks). How big are the gigs your son's group generally plays? I've done outside festivals on split systems, vocals and instruments separate, vocals, drums and the rest of the instruments split, but delay towers getting a general mix, and agree the result is very impressive. Still, it has only been for multi-thousand audiences, never smaller set ups.
You said 'headphone monitors' in one of your posts. Does this mean they use in ear monitoring? You'll be surprised at how critical that mix is; in fact, getting that right from FOH is considerably more of a challenge for the guy mixing than separate speaker feeds.
Good question. I checked local component specialists, and they had nothing in audio range. Perhaps I thought your transformer specialists could wind them for you (as Haufe in Germany did for us many moons ago)
Custom winding will almost always cost more than buying a standard unit, for any reasonable purchase size, unless you're friends with the factory owner There are other manufacturers of good transformers out there, but they are unlikely to be cheap; there are reasons why they are expensive, and nowadays you're unlikely to come across a batch of "excess to requirements, or break down old gear for second hand ones (both of which I have done in the past. But the input transformer is probably the single most critical element in the sound quality, when you're talking about this sort of circuit.
Your average microphone is quoted with the recommended impedance it should feed into; it's own impedance is frequently much lower than this, especially in the low frequencies (dynamic mics because they're heavily inductive sources, condensor because they're active feedback amplifiers). This damps down and drives the lowering impedance on the transformer primary.
When you put a balanced L pad attenuator in front of the transformer, the complex impedance is simplified to a near pure resistance, and you don't get any of this aid. By keeping the resistors as small as possible without loading down the microphone you maintain the damping, if not the supplementary current in the bass region.
Neve, right from the start, used very high quality, expensive and very heavy transformers. Somebody who had started with this philosophy might never have had to learn the compromises of being on a limited budget.
And there is a 1M to the HT supply as well. Since the supply and HT are connected by decoupling capacitors the HT is effectively at ground for ac signals. Hence two 1M resistors in parallel = 500K.
The + input of the op amp is connected to ground to the - input will be very close to 0V. I think it would probably be safe to eliminate C1.
The 100K will will have a small affect the operation of the 50K fader but in practice it should not be noticeable.
You can use a TL081 or 071 in this circuit I think.
P.S I would call VR3 EQ CAL (calibrate the EQ)
Cheers
Ian
Ian,
We worked on the schematic and created the layout (this is the reason for the long delay). Now, we would utilize the Christams for testing the several sections of the circuit and (we hope) to realize the first circuit board. However, we’d like to submit some questions before embarking on these tests. Specifically
1) In the view of your considerations about A262A3E (input transformer), we are re-evaluating this component and we are taking into consideration its substitution with a better one. Sowther seems to have lower prices than Lundahl and their transformers are of good quality (…even if the input impedance has not been reported on the data-sheet). We’d like to ask which of their series (SOWTER PRO AUDIO TRANSFORMERS) represents the best compromise between price and quality?
2) Which value of R would you suggest as a load for the input transformer chosen?
3) How to realize the -20dB pad (you proposed for drums microphones)?
4) We are a little worried about the group of resistors (R21, Vr5 + R1 and R2 of the blue board) in the 2nd-3rd stage (V2-V3). In particular, R12, R1 and R2 (actually of 100K) of the blue board, and Vr5 are in parallel. So, the total R value (seen by V2-C3 and by V3-C9) is about 28K. May that represent a problem?
5) Vr5 (fader) has 1M (of TL082a) and 100K (of TL082b) in parallel to ground. So the real value of fader is 32K. Is that ok or have we change R2 and R3 of TL082b (from the actual 100k to 200-500K)?
6) The output signal we have is about 1200 mV RMS. How we can configure a Pan control (for Right and Left buses) without producing a relevant loss of the signal?
7) If TL082a and TL082b inputs may represent a problem for the V3 stage … is it opportune to put a C in series in order to separate the stages?... on the other hand, TL082a and TL082b drive monitors and Vu-meters, respectively … and for both destinations high quality is not a requirement.
8) As reported on the web-site (VU Meter 3), the vu-meter stage (with IC=TL071) has a “level of entry” of 60-1250 mV. In our case (with TL082b) we have 1218-1234 mV after C9 but higher values are possible too. Can this represent a problem? If so, how to solve it?
9) Are C3 and C4 of blue board necessary?
10) Can C2 and C5 of the blue board be substituted with 1-10 uF (in some schematics are reported 1-10 uF values)?
11) Finally, a cultural explanation. Which is the role of the R (to ground) after an output resistor (often it is omitted in schematics)? … and how to calculate it?
Schematic:
http://agora.lesina.org/?q=node/95
Layout:
http://agora.lesina.org/?q=node/96
We worked on the schematic and created the layout (this is the reason for the long delay). Now, we would utilize the Christams for testing the several sections of the circuit and (we hope) to realize the first circuit board. However, we’d like to submit some questions before embarking on these tests. Specifically
1) In the view of your considerations about A262A3E (input transformer), we are re-evaluating this component and we are taking into consideration its substitution with a better one. Sowther seems to have lower prices than Lundahl and their transformers are of good quality (…even if the input impedance has not been reported on the data-sheet). We’d like to ask which of their series (SOWTER PRO AUDIO TRANSFORMERS) represents the best compromise between price and quality?
2) Which value of R would you suggest as a load for the input transformer chosen?
3) How to realize the -20dB pad (you proposed for drums microphones)?
4) We are a little worried about the group of resistors (R21, Vr5 + R1 and R2 of the blue board) in the 2nd-3rd stage (V2-V3). In particular, R12, R1 and R2 (actually of 100K) of the blue board, and Vr5 are in parallel. So, the total R value (seen by V2-C3 and by V3-C9) is about 28K. May that represent a problem?
5) Vr5 (fader) has 1M (of TL082a) and 100K (of TL082b) in parallel to ground. So the real value of fader is 32K. Is that ok or have we change R2 and R3 of TL082b (from the actual 100k to 200-500K)?
6) The output signal we have is about 1200 mV RMS. How we can configure a Pan control (for Right and Left buses) without producing a relevant loss of the signal?
7) If TL082a and TL082b inputs may represent a problem for the V3 stage … is it opportune to put a C in series in order to separate the stages?... on the other hand, TL082a and TL082b drive monitors and Vu-meters, respectively … and for both destinations high quality is not a requirement.
8) As reported on the web-site (VU Meter 3), the vu-meter stage (with IC=TL071) has a “level of entry” of 60-1250 mV. In our case (with TL082b) we have 1218-1234 mV after C9 but higher values are possible too. Can this represent a problem? If so, how to solve it?
9) Are C3 and C4 of blue board necessary?
10) Can C2 and C5 of the blue board be substituted with 1-10 uF (in some schematics are reported 1-10 uF values)?
11) Finally, a cultural explanation. Which is the role of the R (to ground) after an output resistor (often it is omitted in schematics)? … and how to calculate it?
Schematic:
http://agora.lesina.org/?q=node/95
Layout:
http://agora.lesina.org/?q=node/96
Last edited:
I use Sowter myself and I have always found their quality and sonics to be first class. If you want to retain approximately the same turns ratio I would suggest the Sowter 3195 would be a good choice.
I try to make the secondary load appear as about 1500 ohms in the primary so for a 1:7 transformer the secondary load would be 1500 * 7 * 7 = 73500 ohms. Use a 75K resistor.
See attached sketch.
http://www.ianbell.ukfsn.org/data/20dBpad
No that should not be a problem. The worst case -3dB point will be close to 1Hz and the tube stages should be comfortable driving 28K.
That should be OK. However, I think the direct link between the junction of C9/R21 and the top of VR5 is incorrect??
Sorry, I am not sure what this means. Can you explain a little more please?
The TL082 is a FET input op amp so its bias currents should be near zero. You should not need capacitors on the inputs but you can fit them if you wish.
The LM3915 indicates the level of the voltage on its input pin in 3dB steps so first you need to decide what each LED is going to represent. If we assume +4dBu is your nominal operating level (1228mV rms) then it is common to have the top 2 LEDS indicate +3dB and +6dB above this level and be coloured RED or 2456mV or 3472mV peak. As it stands, the LED VU circuit starts with a unity gain peak detector feeding the LM3915 so an input of 1250mV PEAK or 884mV rms will light the top most LED. So we need a means to reduce the gain so that an input of 3472mV peak lights the top LED. One way to do this is to replace R2 with a 300K preset potentiometer and adjust it so that +4dBu lights up the third LED from the top. This also has the benefit of increasing the input impedance of the LED VU meter and so reduces the load on the fader.
Yes, they are necessary.
10uF for these two is OK.
When equipment is turned on the capacitors are initially uncharged. Output capacitors need a path to ground in order to charge up and this is the purpose of the resistors across outputs.
Good Luck and a Happy Christmas
Cheers
Ian
Ian,
thank you for your clarifications. Is all OK. Only one question and two modifications (due to the availability of the components).
1)
2)
3) Vr3 and R18 has been changed (from 33K to 25K and from 22K to 30K ...becasue 33K lin pot is difficult to find).
4)
5)
Conversely, if we loss part of the signal, we need of a further stage (amplifier stage) after the buses.
In view of this consideration, the question becames "How can we configure a balance control for right and left channels... without inducing a significant reduction of signal level ?"
Now well start with the tests. So we wish us the best wishes for the tests and we wish you the best wishes for Christmas and New year 2011.
thank you for your clarifications. Is all OK. Only one question and two modifications (due to the availability of the components).
1)
Do you mean ... to insert the -20dB PAD circuit before the input mic?
2)
We believe you have an eagle eye too (non only a clinical eye). You are right; it was a mistake on the schematic.
3) Vr3 and R18 has been changed (from 33K to 25K and from 22K to 30K ...becasue 33K lin pot is difficult to find).
4)
Is it the same if we change the 300K trim with a 500K trim (...the 300K trim is difficult to find).
5)
We'll try... We should have a +4dBu signal (1228 mV rms) at the output of our circuit (after V4/C13). If we preserve this value into the 4 buses (vocal, bass, right, left) ... we need of only two further tube stages after the buses: the mixer stage and the output stage (that should be able to drive 50 mt cables !).
Conversely, if we loss part of the signal, we need of a further stage (amplifier stage) after the buses.
In view of this consideration, the question becames "How can we configure a balance control for right and left channels... without inducing a significant reduction of signal level ?"
Now well start with the tests. So we wish us the best wishes for the tests and we wish you the best wishes for Christmas and New year 2011.
Last edited:
Yes, that is where it should go - between the mic input and the transformer.
I spent many years reviewing other people's designs so it is something I do more or less automatically now.
Yes, that should be OK.
That should be OK.
OK, I understand now. You are using the single potentiometer pan circuit which inevitably has a loss of about 6dB. However, you have not said how you are going to do your mixing. Is it passive or will you be using virtual earth mixing? It makes a difference to the possible solutions to the problem.
Cheers
Ian
We think that passive mixing rises several problems (...the voltage from each channel is attenuated by the number of channels plus one ... interaction and crosstalk ...).
So we believe that virtual earth mixing should be better. We red a paper on this topic (The Tube CAD Journal: Vacuum tube mixers) and the schematic "feedback simple tube mixer" seemed simple and appropriate.
What do you suggest ?
OK, that circuit provides unity gain as it stands. For the outputs from the pan pot all you need to do is ensure the virtual earth mixer has enough gain to make up for the loss in the pan pot i.e about 6dB of gain, which you can do simply by doubling the size of the feedback resistor.
Cheers
Ian
If I (Antonio) have correctly understood ... we should use a circuit as that of tubecad for the vocal and bass sections; while, we should use a modified one (changing the cathode 200 ohm resistor with a 400 ohm resistor) for both right and left sections ... and inserting a pan-pot section before these stages (between left and right buses).
In several old tube schematics, I found the input resistors (of mixing stages) to be higher (100k-1M) than those reported in tubecad schematic (10K).
1) Do you think it is appropriate to increase the value of the input resistors (...in order to reduce the interference between the channels)?
2) Which values do you suggest for the cathode capacitor?
3) Could we use a 6DJ8/ECC88 in substitution of 6922 ? ... and...what about a 6DJ8 SRPP stage in this position?
4) Finally ... how to insert the master volume (for each sections ...vocal, bass, right, and left)?
Last edited:
For virtual earth mixing you will be using the feedback simple tube mixer in the TubeCad Journal. This is the tube equivalent of the simple IC op-amp mixer shown on the same page. The gain, to a first approximation, is given by the ratio of the feedback resistor (the 10K from the output to the grid) divided by the input resistor. As these are both 10K the gain is unity. To double the gain you double the feedback resistor to 20K.
Increasing the resistors will not reduce the interference between channels if the virtual earth is doing its job. However, you need to ensure that the mix resistors do not load the pan pots (which will distort the pan pot law) or faders so increasing them is necessary. I would suggest a minimum of 100K.
220uF 63V or something similar.
The 6922 is just a military specification 6DJ8/ECC88 so this substitution is OK. An SRPP stage would probably work just as well and have a lower output impedance.
However, I should warn you about one potential problem with this circuit. The TubeCad Journal fails to mention that the 'quality' of the virtual earth is only as good as the open loop gain of the mix amp. With the values shown for the simple feedback tube mixer the open loop gain is only about 17. The 'value' of the virtual is approximately the feedback resistor divided by the open loop gain. If we have 100K mix resistors and a 200K feedback resistor for our 6dB gain then the 'value' of the virtual earth is approximately 200K/17 = 11.8K.
Now your mixer has 10 sources so if we used passive mixing with 100K resistors, the bus impedance would be 100K/10 = 10K so in terms of isolation we gain nothing by using a virtual earth. To obtain the full benefits of a virtual earth you need much higher open loop gain. If we change from an ECC88 to a 12AX7 (ECC83) we can obtain an open loop gain of around 50 which with a 200K feedback resistor makes the value of the virtual earth 200K/50 = 4K. Not very low but a definite improvement over passive mixing.
Now you see why virtual earth mixing is popular with ICs. They can easily achieve an open loop gain of 1000 or more which would give a virtual earth value in our example of a mere 200 ohms which gives excellent isolation.
The simplest way to do this is to insert them immediately after the mix amplifier and before the output stage.
Cheers
Ian
...uhhhmmm... now the things get complicated.
I clearly understood your "lesson" on the virtual earth mixing. The only thing not clear is the rule to calculate the open-loop gain of a tube.
In any case ... It seems the the main problem is represnted by the faders and balance pan-pots upstream the tube mix... because the mix resistors can distort the pan-law and the fader function (see figure 1).
Indeed, as to faders ... in all the channels they are positioned before the output stage (V4); so they should be unaffected by the summing stage.
As to the balance control ... the problem exists only for right and left output sections (where a balance control is necessary) BUT not for bass and vocal sections (where there is no balance control). So, in the last cases (bass and vocal) we could also consider the utilization of 10K R mix and 10K R feedback...obtaining a gain of 1 and a virtual earth value of 588 ohm (figure 2).
The problem, however, still persists with left and right sections. Also in this case, however, a possible solution could be the use of two separate and independent pots (for right and left channels) as balance control ... utilizing the R to ground (1M) of the previous stage (V4) (see figure 3). So, we could obtain a gain o 6 dB with a virtual earth value of 1176 ohm.
Ian, what do you thing about these considerations?
Is the 1M R to ground (after the output capacitor of 1 uF) necessary in figure 2 and 3 ?
Are 588 and 1176 virtual good earth values ?
I clearly understood your "lesson" on the virtual earth mixing. The only thing not clear is the rule to calculate the open-loop gain of a tube.
In any case ... It seems the the main problem is represnted by the faders and balance pan-pots upstream the tube mix... because the mix resistors can distort the pan-law and the fader function (see figure 1).
Indeed, as to faders ... in all the channels they are positioned before the output stage (V4); so they should be unaffected by the summing stage.
As to the balance control ... the problem exists only for right and left output sections (where a balance control is necessary) BUT not for bass and vocal sections (where there is no balance control). So, in the last cases (bass and vocal) we could also consider the utilization of 10K R mix and 10K R feedback...obtaining a gain of 1 and a virtual earth value of 588 ohm (figure 2).
The problem, however, still persists with left and right sections. Also in this case, however, a possible solution could be the use of two separate and independent pots (for right and left channels) as balance control ... utilizing the R to ground (1M) of the previous stage (V4) (see figure 3). So, we could obtain a gain o 6 dB with a virtual earth value of 1176 ohm.
Ian, what do you thing about these considerations?
Is the 1M R to ground (after the output capacitor of 1 uF) necessary in figure 2 and 3 ?
Are 588 and 1176 virtual good earth values ?
Last edited:
Designing electronics is not easy. It can be almost as complex as being a doctor! You really need read an introductory text on tube circuit analysis. Check around the net and you will find places that explain this. In simple terms a triode can be considered as a voltage generator (equal to mu times the grid to cathode ac voltage) in series with a resistor (the tube's anode resistance). The load (the plate resistance in parallel with the load connected across the tube) forms a potential divider with the tube's anode resistance. So the gain of a typical common cathode circuit is approximately:
Mu * RL/ra where RL is the load and ra is the tube's anode resistance.
In the case of the simple feedback mixer, this gain is also the open loop gain. As a simple rule of thumb, it is rare to get much better than Mu/2 gain from a single common cathode stage so in the case of a 6DJ8 the open loop gain in our example is about 33/2 or around 17.
Correct.
The important thing is by how many times the open loop gain reduces the feedback resistor rather than the actual number of ohms and this is determined by the open loop gain not by the value of input and feedback resistors used. So a unity gain virtual earth mixer with 100K input and 100K feedback resistors is just as good as one with 10K input an feedback resistors using the same amplifier with the same open loop gain. (actually it is slightly better because the feedback resistor is part of the load and as we saw at the start the open loop gain of a simple tube stage depends on the load).
The other thing you have to remember is that the input resistors are the load on the circuit driving them. There is no point placing a heavier than necessary load on the tubes so I would be inclined to use much higher than 10K resistors; as I said earlier I think 100K would be fine. The 'value' of the virtual earth would be 10 times bigger but the input resistors are 10 time bigger too so the isolation is the same. That takes care of the bass and vocal channels.
For left and right you have to also think about the mix input resistors' effect on the pan pot law whether you use one or two pots. A typical two pot pan control uses linear law pots with a 'pull up' resistor from the wiper to the top of the pot equal to half the pot resistance. So in your fig.3 each 1Meg pot would have a 500K from its wiper to the top so that at the centre position each pot provides an attenuation of about 2/3 or 3.5dB. Unfortunately this is completely overridden by the 10K mix resistor which is effectively connected from the wiper to ground.
To solve this I would recommend using 100K dual linear pots with 47K pull ups and bus resistors of 150K and a feedback resistor of 150K. There is no need for 6dB of gain with this type of pan pot.
You do not need it is figs 1 and 3 as the pots do the job for you. I would include it in fig 2 just in case the bus resistor gets disconnected.
See above.
Cheers
Ian
Ian,
please have a look at the schematic of the 4 summing amplifiers. It seems we have respected your suggestions.
Some doubts are ...
1) As to vocal and bass sections (position 1 and 2 of switch S3)... in the worst condition (all 10 channels put on the same position, e.g. all S3 selecting vocal), we have 100Kohm to ground after the output capacitor C13 (1 uF).
Conversely, as to right and left section (position 3 of switch S6) ... in the worst condition (balance pot of all 10 channels put to zero) we have only 3.2 Kohm to ground after C13. Can this represent a problem ?
2) Can S6 produce disturbing noise when switching?
3) Is C13 still correct (1 uF) ?
4) Where to insert the mix resistor (100k and 150K) ?...near the summing tube or near the pot?
5) Is the SRPP summing amplifier correctly designed? Do you suggest it ? ... or do you believe that a simple summing stage is the same ? ... which choice in your hypothetical mixer ?
6) Is the mute correct ?
please have a look at the schematic of the 4 summing amplifiers. It seems we have respected your suggestions.
Some doubts are ...
1) As to vocal and bass sections (position 1 and 2 of switch S3)... in the worst condition (all 10 channels put on the same position, e.g. all S3 selecting vocal), we have 100Kohm to ground after the output capacitor C13 (1 uF).
Conversely, as to right and left section (position 3 of switch S6) ... in the worst condition (balance pot of all 10 channels put to zero) we have only 3.2 Kohm to ground after C13. Can this represent a problem ?
2) Can S6 produce disturbing noise when switching?
3) Is C13 still correct (1 uF) ?
4) Where to insert the mix resistor (100k and 150K) ?...near the summing tube or near the pot?
5) Is the SRPP summing amplifier correctly designed? Do you suggest it ? ... or do you believe that a simple summing stage is the same ? ... which choice in your hypothetical mixer ?
6) Is the mute correct ?
I make it about 32K ohm which is fine. How did you get 3.2K ohm?
Yes. A 1Meg from the wiper of S3 to ground should cure it. By the way you do not need R31 and R32.
Yes.
Near the pot.
Yes, that is exactly as we discussed. The open loop gain will be about 17 because the gain of an SRPP is mu/2.
In my mixer designs I use passive mixing.
Yes, it is correct.
Cheers
Ian
If we consider a single channel (e.g. the first of the 10 channels of the mixer) with S7 switched on "Right" bus and consider also the wiper of Vr6 being in the lowest position (i.e., to ground) the total R is 32Kohm (parallel of 100K and 47k). Now, if all the 10 channels of the mixer have S7 switched on "Right" bus and Vr6 in the lowest position...we have a parallel of ten 32K resistors ... that is 3.2K ohm.
S6 was a wrong reference. The exact component was S7! Neverthless... I think you have correctly interpreted the question. We inserted 1MR between S6-S7 and ground.
To be honest the 1M resistors are reported into the original Aikido schematic. We have not understood their utilities.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.