Discrete DI-Box

[ulysses]

[quote author="mikep"]
has anyone tried making a DI that is basically a DIY version of a famous mic head amp?[/quote]

I've got some prototypes around for a DI based on the "Schoeps" circuit. I bought one of Scott Dorsey's boards to mod an MXL mike but found my V57m already had that circuit as stock. So I put a 1/4" jack on it instead. Since then I've made my own circuitboard that's a whole lot smaller. I just haven't gotten around to getting real metal done to sell it. I've had this thing kicking around for probably 5 years now, maybe I should finish it. So at the moment I would say it's a DIY project. But the point is to fit it into a barrel connector that can be plugged directly into an XLR connector. In order to make that work, it needs some non-DIY-able metal and/or some SMT components.

 

[Samuel Groner]

Thanks for the extensive discussion--been absent for a while because of some important recordings.

This circuit was not design for a given application, so I cannot give specific answers e.g. regarding required headroom or "sound".

I like some of the suggestions by PRR (D1 and D2 should go before the input coupling cap). Maybe we could discuss why drawing current from the center tap should be a bad thing. I realise that if the windings are less than perfectly matched increased distortion may result but I would have assumed that for a Lundahl things are good enough (I might contact the company for their opinion on that). I dislike the resistors as they will eat headroom (i.e. cause lower supply voltage) and increase loading.

Samuel

 

[ulysses]

I like pulling power off the center tap too, if the two halves are well matched - they need to have not only the same number of turns, but also the same length of wire. That's easy or difficult depending on how the transformer is wound. So you have the usual trade-off involving a more elegant and effective solution that relies on a more critical, expensive component. That's engineering for you.

I think it's worth a discussion to establish exactly what the requirements are of a DI. I personally think passive bass pickups are the primary source that will be plugged into a typical DI, and I think passive bass pickups should not see a load much less than 1M ohm unless you're trying to drastically alter the tone. Since many amps have an input impedance around 1M, I'm in favor of making the input impedance as high as possible (10M?) on any DI that may be fed in parallel with a bass amp. So what's the peak output voltage of an unloaded passive bass pickup? It's higher than you might think. Sometimes on the order of several volts, for hot pickups being spanked. On the other hand, if it's loaded significantly (like with a 10:1 transformer DI feeding a 1300-ohm mike preamp input) the peak voltage is significantly lower.
The other end of the spectrum is a low-impedance unbalanced output from an electronic source like a keyboard, drum machine, or other synthesizer. These tend to get plugged into DIs too because people want to make a balanced signal out of them, and also because they tend to need a small amount of gain to get them up to +4dBm line levels. I would argue it's worthwhile to have a different device for this job, but you can consider it a possible application for your DI circuit.
And of course there are a bunch of things in between - guitar pickups, active bass pickups, high-impedance electronic instrument outputs (like the preamp out from a tube amp), guitar effects pedals, and so on. The signal may vary as widely in dynamic range as microphone signals do (or even more so), and the source impedance may vary wildly as well.

 

[bcarso]

Would that be several volts RMS?

 

[clintrubber]

[quote author="ulysses"]So what's the peak output voltage of an unloaded passive bass pickup?[/quote]
Let's not risk anything here. A grown-up bass player in good condition can generate many a Volt on a good day.

I assume a pick will generate the highest peaks (let's ignore slapping-styles since who cares about these anymore these days :twisted: )

For grown-up female bass players some info is here:
[quote author="[url=http://www.carolkaye.com/www/education/tips51.htm]Carol Kaye[/url]"]Leo Fender put me on his oscilloscope and was amazed as the impact of my sound totally knocked his needle off the graph (he said that's never happened before).[/quote]
Too bad (1) she doesn't give a peak-Voltage and
(2) we can't have a glance at Leo's scope that seems to have a needle :wink:

So some while ago I connected some passive basses to a scope as well.
While I couldn't find back any notes I vaguely recall that it'd be wise to
keep at least a 10.. 12 V 'input window' (so 3.5 ... 4.5 V_rms).

(Indeed, 9V powered stompboxes are not always a good idea :wink: )

Bye,

Peter

 

[jrmintz]

Well, you guys got me intrigued, so I hooked my old Fender Bass up to a DMM. On a low E, if I really hit it hard I can get up to about 125 mV. More normal playing is around 50-75mV. That's with both pickups and tone full up.

A grown-up bass player in good condition can generate many a Volt on a good day.

I may not generate as much electricity as I used to, but I got a few left in me. :green: :thumb:

 

[bcarso]

12V p-p... o.k., that's a decent target.

EDIT: although does sound like overkill, having just read Seth's post.

The reason I ask is that the bootstrapping techniques I alluded to tend to limit headroom a bit.

 

[JohnRoberts]

[quote author="clintrubber"][quote author="ulysses"]So what's the peak output voltage of an unloaded passive bass pickup?[/quote]
Let's not risk anything here. A grown-up bass player in good condition can generate many a Volt on a good day.

I assume a pick will generate the highest peaks (let's ignore slapping-styles since who cares about these anymore these days :twisted: )

For grown-up female bass players some info is here:
[quote author="[url=http://www.carolkaye.com/www/education/tips51.htm]Carol Kaye[/url]"]Leo Fender put me on his oscilloscope and was amazed as the impact of my sound totally knocked his needle off the graph (he said that's never happened before).[/quote]
Too bad (1) she doesn't give a peak-Voltage and
(2) we can't have a glance at Leo's scope that seems to have a needle :wink:

So some while ago I connected some passive basses to a scope as well.
While I couldn't find back any notes I vaguely recall that it'd be wise to
keep at least a 10.. 12 V 'input window' (so 3.5 ... 4.5 V_rms).

(Indeed, 9V powered stompboxes are not always a good idea :wink: )

Bye,

Peter[/quote]


Admittedly I researched this almost 15 years ago but at the time I found 9V acceptable for battery power. Recall that the transients from slapping the strings and pick noise will peak several dB above the sustained signal and be a narrow impulse that quickly decays. Clipping such a short duration transient will IMO be difficult to perceive as long as the electronics recover cleanly. FWIW I made my active DB run from a higher rail than 9V when powered from phantom input, or wall wart.

I had the luxury of not even needing to dial an outside line to talk to both amp and axe design engineers, but neither admitted to any interface standard. A little scary, the guitar designers wanted to make the output as hot as possible, while the amp designers cursed them under their breath. 9v batteries are a defacto supply voltage for stomp boxes and active electronics built into guitars, so it seems that a sensible pickup designer would try to target that as a nominal rail voltage, but since when was R&R sensible?

In a phantom powered DB the practical limit for rail voltage is current draw loading down the standard 6.81K supply resistors. I found old school TL07x series easily compatible with that. Today using modern high performance opamp(s) may require doing some current draw calculations. I advocate taking advantage of phantom supply. I even made it independent of the on/off power switch. Batteries always go dead just when you need them, and who always remembers to turn every power switch on or off in the studio.

JR

 

[clintrubber]

The vast differences between those values from Seth & from me will mainly be because of different measuring methods.

What I did was to connect a passive bass directly to a scope (load: 1MOhm//20pF + cable capacitance) and look for the peaks. I expect the DMM-method won't catch these; unless set to peak+hold and then still the measuring-rate might cause missing some peaks.

So what I did was determining a really worst case value; (on a good day :wink: ) those peaks are unsurprisingly really the highest part of the total 'ADSR-sequence'.
I expect that the values from Seth are measured during the sustaining part of the note, right ?

We could discuss whether or not we want to keep those short initial peaks undistorted. IMHO: if it could be done then why not. But it only makes sense of course if we could keep it that way downstream, at least until a fast limiter squashes it in some more behaved way than rude clipping would do.

Bye,

Peter

 

[clintrubber]

[quote author="JohnRoberts"]Admittedly I researched this almost 15 years ago but at the time I found 9V acceptable for battery power.[/quote]FWIW, the highest peaks here came from a passive bass of 1974 :thumb:

Recall that the transients from slapping the strings and pick noise will peak several dB above the sustained signal and be a narrow impulse that quickly decays. Clipping such a short duration transient will IMO be difficult to perceive as long as the electronics recover cleanly. FWIW I made my active DB run from a higher rail than 9V when powered from phantom input, or wall wart.

I had the luxury of not even needing to dial an outside line to talk to both amp and axe design engineers, but neither admitted to any interface standard. A little scary, the guitar designers wanted to make the output as hot as possible, while the amp designers cursed them under their breath. 9v batteries are a defacto supply voltage for stomp boxes and active electronics built into guitars, so it seems that a sensible pickup designer would try to target that as a nominal rail voltage, but since when was R&R sensible?

I fully agree, even while those short peaks might be disturbed @ extreme picking, nobody might notice. So 9V will still do fine.

People do report noticable differences in some cases though when they throw in another 9V battery (like in some active basses & the F*lltone Bass-driver stompbox etc). The differences in sound @ 18V may be caused further down the circuitry though, where things like gain (Bass-driver) & peaking LPF's (Alembic) take place.

Regards,

Peter

I had the luxury of not even needing to dial an outside line to talk to both amp and axe design engineers,

Hmm, a company making both gtrs & amps... did Beh. already exist then ? :wink:

 

[ruffrecords]

Personally I cannot see the point of the transformer, unless it is for a particular sound and it seems an expensive way of providing phantom power. It provides no isolation in the topology used and that's an important property of a DI box.

Or have I missed something?

Ian

 

[JohnRoberts]

[quote author="clintrubber"]
Hmm, a company making both gtrs & amps... did Beh. already exist then ? :wink:[/quote]

Yes, but they were only making rack efx like single ended NR and comps back then...(in Germany)

.... and you would need to dial the pyschic connection to talk to an amp or gutar design engineer there.

JR

 

[clintrubber]

[quote author="JohnRoberts"]Yes, but they were only making rack efx like single ended NR and comps back then...(in Germany)

.... and you would need to dial the pyschic connection to talk to an amp or gutar design engineer there.

JR[/quote]
It was actually a bit of a lame joke on my side... unnecessary since we have already something going in the Brewery right now.

Let's quickly switch back again to designs'n'peaks'n'levels, interesting thread.

Regards,

Peter

 

[ulysses]

[quote author="ruffrecords"]Personally I cannot see the point of the transformer, unless it is for a particular sound and it seems an expensive way of providing phantom power. It provides no isolation in the topology used and that's an important property of a DI box.[/quote]

You have a good point about this particular DI design - if you can't have the isolation, then why bother with the transformer?

On the other hand, on the topic that has been discussed on the last page or so of posts, the traditional passive DI which contains a transformer and nothing else also does a good job of solving the problem of the rather large transient peak voltages coming off of a passive bass pickup. The result is a nicely rounded-off peak that's substantially smaller than the original transient and closer to being in line with the average signal level. Transformers recover from overload much more neatly than transistors usually do.

I wonder if this might be the reason that passive DIs have been favored, despite their other obvious flaws (expense, bulk, and inability to prevent excessive loading of the pickup).

But if you have active electronics in front of the transformer that clip on peaks and don't sound good doing it, then you're left with no advantages at all over a transformerless design.

Just make sure anything in front of the transformer has lots of headroom - I think 10V peak (20V p-p) might be overkill, but then again it might not.
And what about active pickups? Maybe they can be ignored like the bass slappers?

 

[PRR]

Most voltmeters are useless on transient audio.

History tells the output levels of guitars. Early ones sometimes used grid-leak bias and overloaded at 0.6V peak. But the end of the 1950s, Fender had switched to 12AY7 and 12AX7/7025 inputs cathode-biased to take up to 1.5V peak. Then he added the 2-jack scheme which offers 2:1 attenuation on one port (though many players are unaware of that function). That plan has stood the test of time and has been universally copied.

One outfit which resistted stealing Fender's designs, Ampeg, still arrived at about 1.2V peak straight and a second input with substantial LF attenuation.

So it seems clear that 0.6Vpk may not be enough, 1.5Vpk is often enough, but you may meet 3Vpk. Or in RMS: up to 1V or 2V.

Fairly simple circuitry will handle 1Vrms off a 9V battery, 2Vrms (6Vpp) is a little tougher.

The proposed 24V supply seems plenty ample, justified only because we have 48V on tap.

And because, in my experience, a DI often turns into a general purpose 1/4" input, and may be fed lines and even speaker levels.

At the other extreme: a typical sensitivity rating is 20mV for Full Power. This does not mean anybody plays 20mV, more like they should be able to play with ~50mV peaks pushing the amp into overload. This may be easier picking for better stamina, or the guitarist may have to turn-down his guitar volume pot to balance cable capacitance against pickup inductance and HF peaking. Still you'd be wise to design so that 20mV is far-far above your input noise level (usually not a problem).

> whether or not we want to keep those short initial peaks undistorted.

Well, capture without injury.

In fact we DO want to bend them a bit. I believe the reason Leo didn't go over to truly high-level inputs is that guitarists WANT 5% THD and some clipping at maximum effort. That lack of preamp bending is why chippy inputs sound boring. But such sound-smashing is not the job of the DI. You can semi-easily tap further up inside the amplifier, or its output, or mike the speaker; and this is generally done. Half the sound of lectric gitar is in the amp and speaker. But when forced through tape loss, we sometimes want less amp-sound, so we mix amp-mike with DI to taste (preferably the next day, not on the spot).

The DI should just be clean.

It should take at least 2Vrms input, and more is a bonus.

It should be a "high" impedance, but pickup peaks are OTOO 300K, so 1Meg is generally ample (and transformers are compromises).

I agree with Ian that galvanic isolation is a great benefit. Ideally we are all on one big happy clean ground; in real life sometimes the best-sounding amps are electrical mongrels. This is also a problem on location, where you don't own the electric system, it was installed lowest-bidder, maybe by several contractors over several decades, and you need to record on the far side of the building from the stage.

 

[Samuel Groner]

You have a good point about this particular DI design--if you can't have the isolation, then why bother with the transformer?

The main reason to have a (step-down) transformer here is to transform the load to a level where the output stage can easily run in class A and to provide a phantom power draw off point without additional voltage drop (BTW, Per Lundahl agrees with me that there should be no disadvantage by drawing the current from the center tap). Even though the center tap is AC- grounded, the output transformer still provides very good output CMRR. Galvanic isolation is a nice thing to have, but if you get more than 1 Vrms common mode signal in your studio you have a serious problem which should IMO be treated by prober installation rather than by galvanic isolation of audio signals. Live use is another thing, obviously.

Galvanic isolation would be possible by the addition of a DC-DC converter, I'd say.

Samuel

 

[Samuel Groner]

I updated the schematic linked on the first page to incorporate several changes suggested in this thread. In addition to this I changed the input transistor to 2N5457 for lower input capacity and the transformer to LL1576 (according to the recommendation of Per Lundahl).

I simulated some bootstrapping cofigurations for the input transistor but I got stability problems with high source resistances--any reason or solution for this?

Samuel

 

[ulysses]

[quote author="PRR"]I believe the reason Leo didn't go over to truly high-level inputs is that guitarists WANT 5% THD and some clipping at maximum effort.[/quote]
This may be getting a bit off-topic, but I believe that at the time Leo borrowed his amp designs, guitarists did not want distortion. Rock and Roll was invented as a means of making something worthwhile out of the amplifier limitations they were stuck with at the time. If the first amps had been clean, or had some other kind of distortion instead, we'd be sitting on an entirely different musical history based on some other kind of sound. As it happens, Fender-style amp distortion has a wonderful tendency to synthesize the 3rd harmonic out of thin air, which paved the way for untrained guitarists to sounds like musicians because they only needed two fingers instead of three to play barre chords.

 

[Guest]

Fender-style amp distortions can't be viewed without fender-style wooden axes with steel wires on them in the magnetic field. Leo clearly knew what he was doing, he took resonating deck out of a gutar and made it remote, i.e. speaker cabs and amps were parts of guitars that did not have resonating decks, they were dumb and could not sound good without proper resonating deck that was called from now "Speaker Cab".
Acoustic and semi-acoustic guitars never needed such distortions and colorations, they were self-containing instruments and used pickups and amplifiers only to amplify their own sound, without modifications.
For me, DI should not substitute a gutar amp, nor any other part of guitar. It must take full signal "as is", without modifications, from the signal source.

 

[bcarso]

[quote author="Samuel Groner"]I simulated some bootstrapping cofigurations for the input transistor but I got stability problems with high source resistances--any reason or solution for this?

Samuel[/quote]

Typically there is a negative input Z at some frequency. A small (few pF) cap to common is usually sufficient (gate to common that is).