REDDI DI

[alexc]

In the scenario where a single ended config is used, and both halves of the tube are needed in parallel to drive the se transformer with sufficient current, a 'level' control could be done with a t-pad on the output transformer secondary.

Like a 600 ohm balanced T-pad as used by Hairball's 1176 kits or Classic API kits.

I don't much like putting a pot at the input, unbuffered into the line amp. Although I could be misinformed
And this way, you can drive the transformer hard independent of the desired output level.

For an SE line amp, I think this is a good route to some interesting .. 'voicings'

Of course, there are many other configs which would also serve.

I just like that one

 

[abbey road d enfer]

In the scenario where a single ended config is used, and both halves of the tube are needed in parallel to drive the se transformer with sufficient current, a 'level' control could be done with a t-pad on the output transformer secondary.

Which is OK as long as you don't overload the input...

And this way, you can drive the transformer hard independent of the desired output level.

Driving hard is good up to a point where it's driving too hard...

 

[alexc]

Yes, driving the input grid too hard could be a problem - so providing a bias of a couple of volts so should provide plenty of headroom for even the strongest input signals.

Maybe a sensitivity switch or even hi-lo input jacks to provide some flexibility.
Or a potentiometer if one wants to.

Lots of ways to go.

 

[abbey road d enfer]

Yes, driving the input grid too hard could be a problem - so providing a bias of a couple of volts so should provide plenty of headroom for even the strongest input signals.

This is a very narrow view of the problem. Increasing bias does indeed increase the input admittance, but also considerably reduces the operating current, which in turn decreases the output capability, which results in no improvement in signal handling and increases noise.

 

[Biasrocks]

We can take a cue from the Ampeg schematic, a 1M volume control is between the two tube stages (see attachment). Since we only need about 16db of adjustment it would have to be modified slightly.

We would have to optimize the input impedance for around 1M and have the output capable of driving a ~ 5K:600 output transformer.

Regards,
Mark

 

[dmp]

For the design posted by Bernbrue can an output transformer be used to get a balanced out? 
What to look for in buying 6N1P tubes? Seems the sellers on ebay are all eastern European and I can't make sense of any branding on the tubes..

So I've recorded a lot of guitar and bass using NYD's 1 bottle as a DI, re-amped into a 70s B15, through a Showman 15" cab. Sounds fantastic. Making a simple tube DI in one of dandeurloo's tube PSU boxes sounds really appealing.
And my opinion - I wouldn't think you could get a 'B15' sound with just the first tube stage for a DI - and it would make more sense to have the tone stage on reamping, not on recording the DI.

 

[Biasrocks]

And my opinion - I wouldn't think you could get a 'B15' sound with just the first tube stage for a DI - and it would make more sense to have the tone stage on reamping, not on recording the DI.

Agreed.

Regards,
Mark

 

[Michael Tibes]

This schematic should be a good inspiration for a very good sounding DI, but it might contain errors. Open question is the output transformer and the out level pots value. I'm pretty sure about the out pot's location, though it doesn't seem to be used in an obvious way to me. Maybe that's part of the magic? Or I need new glasses... Out transformers primary dc resistance is about 950R. Anyone on the transformer?

Michael

 

[CJ]

that tube would be happy with a 2 k transformer primary which we could step down easily to 500 ohms, (easy math)

2000/500= 4 impedance ratio

turns ratio is sq root of 4 which is.....





2,   


so we need a 2:1 transformer with 2k pri z, which handles 20 ma dc,

good news is we generate low ac flux because the output voltage can only be about 20 volts unless somebody overdrives the tube.

you could even go 1:1 as now a days, most stuff could be driven with a 2k secondary,

freq response -high end is cake walk with 2:1 turns,

bottom end needs scrutiny as we have a dc bias,


pick a number between 1 and 20, say 10.

design for 10 hz, 2K z input, 10 ma dc, 20 volts ac, 500 ohms out,

typical 500 ohm output winding is 5 to 20 henries, lets use 20 henries for good bass,

so now we figure pri Z, 4 to 1 ratio, so pri Z = 4*20=80 henries,

we can guess at the size, a 12at7 takes a small core for the reverb trans, so a 625 EI would be overkill, but we do have 20 ma dc, so 312 EI might be to small, so maybe  start with  a 1/2 inch tongue and see how many turns we need for 80 henries,

50 EI - look up the specs in the Mag Met catalog, all we need is 3 things to compute a SE transformer,

cross section, Henries coeef, and MPL-mag. path length

ok i will be right back...

 

[CJ]

50 EI did not cut the mustard, so 625 EI it is,

remember, this is good to 10 hz even with the dc,

3600 turns with an .005 in gap is the ticket,

i use inches for the gap as that is how the paper is cut over here,

a xfmr that handles 20 hz would be about half this size, so if this is for guitar only, then a smaller core would work fine,

50/50 ni could be used also.



see spread sheet for flux and all that>

and if you understand all this, you get to listen to Burning Spear,

live and we got it archived,  Paris 88 of course,

http://www.youtube.com/watch?v=pNIo7B69tps

 

[alexc]

That's pretty much exactly the configuration that I had in mind 

Substitute the tubes, transformers and attenuator of choice and adjust the operating point accordingly.

I would add a variable cathode resistor to allow for tailoring of the bias for different pickups amd call it the 'narrow view' control    Possibly even a switched cathode cap bypass for some gain variation.

 

[chrispbass]

Hi all, very intersted in this discussion, especially as a bassist. The design looks very simple judging by the internals photo posted previously, but was wondering if stripping the G9 down (although a 2 tube design) would not give as good or greater results? i'm thinking of doing this P2P as I think one or even two channels could be built reasonably cheaply as the input tx is no longer needed and the input selector, hpf and phase could be omitted, obviously along with the phantom supply. The input gain switch could easily be built on the back of the lorlin switch, which leaves a small amount of passive components left to wire P2P. There is a spare B+ and heater supply pcb combined with the g9 pcb set so that is covered. Perhaps a bit more involved than the REDDI, but would this be a worthwhile pursuit? I estimate about £100 for 2 channels component wise!!
Chris

 

[alexc]

With 2 tubes available, there are many circuit topologies you could use for a nice bass DI.

The g9 with all the things yo mention removed is a 2 gain stages with feedback gain control feeding an srpp+output transformer.

Thats fine although I think there would be a lot of gain there - probably too much, thus requiring a lot of attenuation and/or feedback.  But all doable, although I'm not sure that lots of feedback is needed in a di.

Alternatively, you could do some traditional fender style bass amp preamp sections, like 2 stages as a grounded cathode amp+cathode follower feeding an srpp+transformer.

Or you could do a single gain stage feeding a white cathode which is the other gain stage (to make up traffo stepdown) of the first tube feeding a 2 stage cathode follower + transformer.

I've done both of the above with great result, particularly the first one, for the bass. With an edcor traffo.
I use it all the time. Mine included the tone controls as well, so it has 3 tubes all up.

I would also use a simple psu from the amp building world and dispense with any dual transformer+g9 psu pcb stuff.
Just overly complicated for a straight ahead di build without any real benefit. 

You can do a basic psu job with very low noise without any dc or regulation.

Good luck with it.

 

[evil grill]

Nice stuff!

So how would this unit behave if the cathode-resistor got substituted by 2 IN4148's?
And is the input impedance really 240k? Should'nt the grid resistor (1M) determine that?

 

[abbey road d enfer]

So how would this unit behave if the cathode-resistor got substituted by 2 IN4148's?

I don't think it would change much. If the stage was operated outside of class A, it may make a difference. But here, with constant-current, the load lines are not likely to vary much.
I've done a brief sim; it shows minor gain variation (<0.2dB) and a slightly different LF behaviour because the dynamic resistance of the diodes is smaller than the cathode resistor.

And is the input impedance really 240k?

No.

Should'nt the grid resistor (1M) determine that?

Indeed. However, the input capacitance due to the Miller effect is about 60pF, which computes at less than 150k at 20 kHz.

 

[evil grill]

Thanks for the reply!

I have recently gotten the lust for building stuff again and I need a nice DI, plus I've recently gotten my hands on a bunch of used Lundahl transformers that need to be put to good use.

Please feel free to comment on the schematic.

I don't think it would change much. If the stage was operated outside of class A, it may make a difference. But here, with constant-current, the load lines are not likely to vary much.
I've done a brief sim; it shows minor gain variation (<0.2dB) and a slightly different LF behaviour because the dynamic resistance of the diodes is smaller than the cathode resistor.

Any improvements over the original design or is this detail pointless?

... However, the input capacitance due to the Miller effect is about 60pF, which computes at less than 150k at 20 kHz.

Any suggestions that would help LF linearity?

 

[ruffrecords]

I can't believe I missed this thread until now. My comments on what has been said so far.

The 6N1P is not an equivalent to the 6DJ8/ECC88 no matter what certain manufacturers data sheets may claim. I have tested many of them against the 6DJ8 and I the only specification the two might have in common is their mu. The 6N1P definitely has different gm and rp values and hence its biasing is quite different. I have also measured its intrinsic distortion which I have found to be significantly higher than the 6DJ8/ECC88 but that might be exactly why the REDDI designers chose it. If you do want a Russian equivalent to the 6DJ8 then the 6N23P is pretty close on all measured parameters.

I like the SRPP idea because it gives a nice low source impedance to drive the transformer,and produces nice amounts of mostly second harmonic distortion. Most importantly it avoids having dc going through the transformer which means an un-gapped transformer can be used. This means, for a given core size, the primary inductance can be much larger and hence a much better bass response is obtained. That probably explains why the output transformer in the original is so huge.

The SRPP design posted is possibly a bit low on HT voltage although this does mean you can use any 6DJ8 type and you don't need to elevate heaters. Personally I would raise the cathode resistors to 330 ohms and the HT to 250V. Plate dissipation is a little higher but the voltage swing is greater and distortion a little lower but this is a matter of taste really.

Distortion is definitely a function of load and you will get about twice the distortion with the transformer loaded with 600 ohms as with 10K. You can do a simple output level control with a 1K pot. The Carnhill VTB2291 ungapped transformer makes a nice 2:! transformer, is relatively cheap and has a good primary inductance.

Frequency response at the top end is dominated by Miller effect which in turn depends on the driving source impedance. With a 1 Meg input pot the driving impedance can easily be as high as 250K which gives a -3dB point at a little over 20KHz so I am not sure how they get their claimed 60KHz frequency range with a claimed 1Meg input impedance unless they use a cathode follower feeding a singled ended stage. I am not sure if a single triode in the 6N1P has enough drive capability to get +22dBu into 600 ohms which is 158mW.

Cheers

Ian

 

[tv]

Ian, you're our "Ruffpedia".

 

[evil grill]

...
The SRPP design posted is possibly a bit low on HT voltage although this does mean you can use any 6DJ8 type and you don't need to elevate heaters. Personally I would raise the cathode resistors to 330 ohms and the HT to 250V. Plate dissipation is a little higher but the voltage swing is greater and distortion a little lower but this is a matter of taste really.

Distortion is definitely a function of load and you will get about twice the distortion with the transformer loaded with 600 ohms as with 10K. You can do a simple output level control with a 1K pot. The Carnhill VTB2291 ungapped transformer makes a nice 2:! transformer, is relatively cheap and has a good primary inductance.

Frequency response at the top end is dominated by Miller effect which in turn depends on the driving source impedance. With a 1 Meg input pot the driving impedance can easily be as high as 250K which gives a -3dB point at a little over 20KHz so I am not sure how they get their claimed 60KHz frequency range with a claimed 1Meg input impedance unless they use a cathode follower feeding a singled ended stage. I am not sure if a single triode in the 6N1P has enough drive capability to get +22dBu into 600 ohms which is 158mW.

Cheers

Ian

Nice!

The bunch of Lundahl-transformers I got are all only capable of 1:1 or 2:1, (LL3802) and I got exited to see some use for them in a DI tube-design.  No info on inductance. The company said the info on their site is all they have to offer concerning these transformers.

Correct me if I'm wrong, but I interpret your suggestions as the attached schematic.


PS. Anyone got tips on a good low dropout voltage regulator besides NCP57302 for DC-heat?

 

[dmp]

The 1K pot on as the output level does not look right. Maybe it should be a voltage divider and go to ground?