Bo Hansen's 325 DI: Help me understand the circuit (n00b content)

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

finbase

Well-known member
Joined
Jan 6, 2014
Messages
47
Location
Espoo, Finland
Hello folks!

Yet another DIY-n00b question, so paternal patience is required…  ;)

I'm planning to build one 325 DI for bass (I am a professional bass player), based on Bo Hansen’s schematic. Since I am a beginner, I want to study the 325 thoroughly, until I can really understand what's happening in it, and what the every discrete component and connection actually stands for.

I think I've got a good overall understanding on the circuit this far, but there are a couple things (or actually caps) I don't understand. They're probably crystal clear to you, but I can't find the answer either on the internet or in my books. I’ve tried several hours now, with no success. So I decided to humble myself and ask for help in public… (and since I’m anonymous, there’s not much to lose, is there!  :))

1) Although this might be VERY simplistic question, I just don't understand how the -10db pad works. I can figure out the component layout, but I don't understand what the parallel RC circuit actually does... What happens to the current flow, voltage etc., what's the physical phenomenon when the pad switch is either on or off? I can understand the resistor, but what that parallel capacitor is supposed to do?? (Why oh why?!?)

2) Similar question: I don't understad what the Capacitor C6 in the Op Amp feedback circuit stands for. If I interpret the circuit right (this is a non-inverting amp, right?), the resistor R3 (47K) and the 47k rev. log potentiometers control the amount of the output amplitude. But that parallel RC circuit (R3 and C6) is confusing again. I don't understand WHY there's that capacitor...

Please, if you have time for this, try to explain what those parallel RC circuits are used for in audio circuits such as preamps. I understand it's usage in oscillators, but in the 325 circuit... I'm clueless!!

The schematic is attached below. (If it’s not permitted, please let me know, I’ll remove it.)

Thank you in advance!!

finbase

(PS. hopefully my English is understandable, I'm not a native)
 

Attachments

  • API 325 DI mod..jpg
    API 325 DI mod..jpg
    155.3 KB · Views: 198
You will notice the 1M resistor from the + input of the op amp to ground. Also from the _ input is the 2M2 resistor of the 10dB pad. These two resistors together form a potential divide that have a loss of 1/(1+2.2) times = 0.3125 = 10.1dB. So its a 10dB pad.

The capacitor is very small, 18pF so at low frequencies it has no effect. As the frequency invreases the impedance of the capacitor gets smaller and starts to short out the 2M2 resistor so reducing the loss in the pad. As this happens only at high frequencies it gives a small treble boost. The frequency the boost starts is when the reactance of the 18pF equals 2M2 which turns out to be just over 4KHz.

Cheers

Ian
 
{Ian types faster than me.}

There's capacitance EVERYWHERE.

Both networks are voltage dividers.

Both networks are resistive over the audio range, or most of it.

Both networks use a small physical capacitor to balance stray capacitance.

Input pad: for low frequencies we have 2.2Meg against 1Meg. The output is 1M/3.2M or 0.31 or 10.1dB down.

We see that Bo has added 18pFd across the 2.2Meg.

The corner of 18pFd and 2.2Meg is about 4KHz. So most audio (nearly all bass guitar) is unaffected by the 18pFd, but higher frequencies might be.

If we assume there is about 36pFd or 40pFd of stray capacitance across the 1Meg, (one opamp, a couple of Zeners, and whatever wiring), then the supersonic padding in the capacitor divider is approximately equal to the resistive padding.

If he didn't do the 18pFd, but had ~~36pFd of stray across a ~~1Meg pad, the response would be flat to 4KHz and drop 6dBoct above that point. Guitarists might notice. Synth players using a DI certainly could tell. And geeks who test-bench 20-20K would call it a problem.

The cap on the NFB gain network... first, figure 47pFd across 47K is a corner at 72KHz. Whatever it does, isn't in the audio band. But the opamp has gain far above the audio band. And a roll-off. There's also a lot of stray capacitance in the gain-pot wires. All of these cause phase-shifts. Three R-C phase-shifts is sufficient to invert phase and turn negative feedback into positive feedback at some high frequency. If there's still some gain at that frequency, it WILL oscillate. The efect of the 47pFd in that position is to reduce overall phase shift. (Yes, this gets complicated.)

Anyway, while we want 20KHz response, anything much higher is garbage. Radio waves, static crackles, lamp dimmer hash. While we won't hear it directly, it may intermodulate with other tones and throw difference products into our precious audio. We usually want some explicit roll-off above the audio band.
 
Thanks for your help Guys!

Ian

Ok, I understand the -10dB pad now. It was just a bit confusing, since at first I assumed that the signal passes through that 2M2 resistor all the time... (The schematic's switch symbol is a bit "misleading" for a n00bey like me.)

That 18pf short circuit makes sense also. Clever! :)

PRR
And thanks PRR for your reply also! Well, it's getting complicated indeed...  :D Well, a few years more, and I might understand then!


Thanks!

finbase
 
Just remember that complicated things are made of lots of simple things stuck together. Understand the simple things and you can understand the complicated ones.

@PRR. I usually check through the group over breakfast which is probably about 2am your time so I suspect it's more a case of the time difference works to my advantage rather than exemplary typing skills.

Cheers

Ian

Cheers

Ian
 
PRR said:
The efect of the 47pFd in that position is to reduce overall phase shift. (Yes, this gets complicated.)

I've always wondered:  would it be fair to call this "cap across the feedback resistance" dominant pole compensation, since we're placing this new pole (hopefully) well below where the other stray elements start to add in in their own poles/zeroes?
 
Hi,
just for the sake of learning is it possible to explain the other components Bo added?
that would be great

10 K resistor in series with the input
2x 12V zener diodes across input and audio ground
0.1uf cap, is it to block DC?
2x 1n4148 diodes across the Opamp's inputs
0.047uf cap on the ground lift switch terminals

If someone could shed some light that would be great
thanks
 
PRR said:
http://www.geofex.com/circuits/what_are_all_those_parts_for.htm

Hi PRR, thank you so much for the article, great explanation.

So let's see if I learned something:

- The 10K resistor in series at the input should be part of a radio frequency low pass filter, althoug there's no cap to ground after it.
does it form a filter with the capacitance the 2520 has to ground?

- the 2x 1n4148 diodes, are there to protect the 2520 input from overvoltages, spikes...

- 2x 12V Zener should be also some kind of protection, but I don't know what they do or protect

- The 0.1uf capacitor should set the frequency of the High pass filter in conjunction with R3 47K.
is it also used to block DC fron reaching the 2520 inputs?

Why is not this on the original 325 schematic?
Is it just a better practice or just needed for instrument signals and not for Balanced Line level?

- the 1 Mega resistor from input to ground, I think it sets the input impedance, but on the RG explanation it goes t Rbias and not to ground. Confused here

- The 0.047uf capacitor on the ground lift switch  I don't know.
Is it to filter the ground?
is it to prevent pops or noises while switching from Ground to Lift?


 
The big difference: the original lived in a studio with competent engineers and generally known-strength sources. A "DI" gets ANYthing shoved into it, from weak ukulele to Marshall Amp output.

You are right, the added 10K does little for RF unless there is a cap. I suspect that if there is that much RF on stage, other interfaces will also be interfered,  so RF rejection was not considered.

The diodes "protect" the inputs, but a 100W guitar amp will easily smoke most diodes. Adding 10K limits the current. Say 100 Volts goes in: 100V/10K is 10mA, and even small diodes can handle that.

I've probably missed a point but this gives you a start.

> thank you so much for the article

Not mine; the notorious R.G. Good site.
 
Whoops said:
- The 0.1uf capacitor should set the frequency of the High pass filter in conjunction with R3 47K.
is it also used to block DC fron reaching the 2520 inputs?

Yes.

- the 1 Mega resistor from input to ground, I think it sets the input impedance, but on the RG explanation it goes t Rbias and not to ground. Confused here

On RG the opamp is powered from a single rail.


- The 0.047uf capacitor on the ground lift switch  I don't know.
Is it to filter the ground?
is it to prevent pops or noises while switching from Ground to Lift?

When the switch is on you have both DC and AC ground paths,  but when the switch is off you have only AC ground path.
 
sahib said:
When the switch is on you have both DC and AC ground paths,  but when the switch is off you have only AC ground path.

Thank you so much PRR and Sahib for your shared knowledge.

Sahib, about the Ground switch, what's the practical implication of having an AC Path to ground and not a DC path?
Shouldn't the switch lift lift completely the ground connection to the next machine in the chain?

Thanks
 
In general, in both equipment the 0VDC is tied to the mains safety ground/earth at the power supply. So you already have a DC link through the mains earth between the two equipment. 

When you have a second DC link through the screen (of the signal cable) you then have a complete loop. 

There might be a case where you have to cut this DC path (because the hum/noise currents will circulate in combination with DC currents) but  still need to maintain  the AC path,  not only because of the mains hum, but also the way the other noise currents are interacting.







 
sahib said:
There might be a case where you have to cut this DC path but  still need to maintain  the AC path,  not only because of the mains hum, but also the way the other noise currents are interacting.

Thanks Sahib,
I understand the DC lift, just not when it might be needed to maintain the AC Path. I normally just lift the ground completely, Pin 1 lifting at the audio output
 
It would be difficult to say particularly in multiple interconnection situation.

In this particular application perhaps Bo can tell us what his original intention was, but what I see is that by lifting the switch you are still providing a path for the noise that may be imposed onto the screen at this end of the cable.
 
Whoops said:
Thanks Sahib,
I understand the DC lift, just not when it might be needed to maintain the AC Path. I normally just lift the ground completely, Pin 1 lifting at the audio output

Ok - so for example, if pin 1 is tied to Ground, or PSU 0V in the mixer, and pin 1 is tied to the PSU 0V reference in the DI, and a cable with shields connected to pin 1 on both ends is used to connect the two (which is common practice, tho breaks one Morrison's grounding and shielding fundamentals) - this creates an alternate path back to ground for each power supply and is the closed loop that Sahib is talking about - lifting the DC connection on the shield breaks the DC ground loop but also lifts the reference of shield at one end, which at best - just decreases the shielding effectiveness, and at worst creates an antenna for RF...

a small value cap looks like an open circuit to DC and a short circuit to AC...

so RF or any AC content formed across the shield is shorted to ground thru the cap, but the ohmic connection is broken so there are not multiple paths to ground thru the shield...



 
mutterd said:
Ok - so for example, if pin 1 is tied to Ground, or PSU 0V in the mixer, and pin 1 is tied to the PSU 0V reference in the DI, and a cable with shields connected to pin 1 on both ends is used to connect the two (which is common practice, tho breaks one Morrison's grounding and shielding fundamentals) - this creates an alternate path back to ground for each power supply and is the closed loop that Sahib is talking about - lifting the DC connection on the shield breaks the DC ground loop but also lifts the reference of shield at one end, which at best - just decreases the shielding effectiveness, and at worst creates an antenna for RF...

a small value cap looks like an open circuit to DC and a short circuit to AC...

so RF or any AC content formed across the shield is shorted to ground thru the cap, but the ohmic connection is broken so there are not multiple paths to ground thru the shield...

Thank you for the explanation Mutterd
 
You've got it all explained in an excellent way, can not make it better myself, thank's guy's.

Perhaps unnecessarily, but would still like to explain what I was thinking when I made my modification in all simplicity.

First of all, the resistor R1, 12 K (2520 amp input reference ground) was changed to 1 megohms, to obtain a suitable input impedance for use as a DI box.
The capacitor C5 is removed completely.
Gain potentiometer was selected with the value 47 k reverse logaritmic, as with other resistor values, the minimum gain is 5 dB and the maximum gain of 40 dB.

The instrument/high-Z input circuit has two 12 volts zener diodes to ground, (connected in reverse to each other in series) for taking care of a too strong input signal, that could damage the expensive 2520 amplifier, and the 10 k resistor is to protect the zener diodes.

The capacitor of 0.1 pF is to block  DC voltage which may be exposed on the input.

The two 1N4148 diodes in parallel in opposite directions between +/non inverted and -invertet inputs, (anti-latch-up diodes) is also an extra protection for the expensive 2520 amplifier.

The PAD are made in a simple manner, where the 2.2 megohms resistor, is the series resistance in the voltage divider, and where the 1 mohms resistor for amplifier input reference, are the shunt resistance to ground in the voltage divider.
This becomes -10 dB pad/loss, when not 2.2 megohms resistor is shorted by the switch.
High-resistance voltage divider have problems with flat frequency in the high treble range, for that reason there is a 18pF capacitor in parallel over the 2,2 megohms resistor to compensate for this, so the frequency response becomes flat well above 20 kHz.

The C6, 47 pFcapacitor, are API original, and is located in
parallel across the feedback resistor R3, and this  to make the 2520 amplifier stable and calm, to avoid self oscillation and sensitivity to
the HF/RF interference.

The Ground lift switch is made in the simplest way, as usually used in passive DI boxes where phantom power is not used.
Can best be done with just break up ground from the XLR pin-1 with a switch.
This usually solves hum/ground loop problem, but can cause a high-pitched buzz instead.
To easily solve this, I have connected a 0,047 uF capacitor across the switch, which ensures that the HF /RF interference is always de-coupled to ground.

Finally, the transformer outputs.

The 2503 transformer has three identical secondary coils/output taps, which I used for the three different output alternatives.
Sec. 1,  is a balanced normal 1:1 line output.

Sec. 2,  is a typical balanced DI box output with -30 dB attenuated level for driving a mixing console or microphone preamp input.
If another output level desired, change the 220 ohms resistor up or down in value.

Sec. 3, is a completely isolated ground free output to drive the back-line or monitor amps.
This output is unbalanced and has a variable level with a potentiometer.
Note, the tele-jack must be insulated plastic type, so sleeve/ground not is in contact with other electronics ground system on the 325 pc-card or metal box/case.
If problems with the high-pitched buzz occurs, connect 0,047 uF capasitor from tele-jack sleeve/ground to the ground system on the 325 pc-card.

--Bo
 
Back
Top