REDDI DI

[ruffrecords]

Done some new measurements. Check it out on flickr.
http://www.flickr.com/photos/95158972@N07/

0.775V in
169 VDC - gnd/b+
154 VDC - gnd/pry wire
2.25 VAC - b+/pry wire min pot
13 VAC - b+/pry wire max pot
0.015 VAC - secondary min pot (connected to the circuit)
2.9 VAC - secondary max pot (connected to the circuit)
4.25 VAC - secondary (disconnected from the circuit)
15.6 VAC - b+/pry wire ( secondary disconnected from the circuit)
Voltage on the 220R = 1.84V DC

That makes more sense - virtually no output with the pot at min.

Open circuit output of 4.25V rms. 1K loaded output of 2.9V rms which means there must be 4.25-2.9 = 1.35 V rms across the secondary output impedance which in turn means the output impedance must look like 1K x 1.35/2.9 = 465 ohms.  15.6V rms on the primary with 4.25V rms open circuit on the secondary implies a turns ratio of 3.67:1 which in turn implies the plate circuit must look like a source impedance of about 6265 ohms

1.84V across 220 ohms means idle current is about  8.4mA which is also about right for this type of triode. Total idle current of 16.8mA drops 15 volts across transformer primary which means its dc resistance is about 890 ohms

1K reflected to primary is 13.5K. Stage gain is therefore mu x 13.5/(13.5+6.265+0.89) which is about 20 times or 26dB which is confirmed by 15.6Voutput from 0.775V rms input (26dB gain)

Cheers

Ian

 

[dirty1_1garry]

thanks Larry!

I'm Garry, not Larry  8) 
;D

 

[CJ]

i meant your other brother Larry, Garry,  ;D

or dirty harry, which ever comes first,

 

[Hank Dussen]

By the way, how is the 'thru' implemented?
Is it just a parallel input?

 

[dirty1_1garry]

By the way, how is the 'thru' implemented?
Is it just a parallel input?

Exact just a parallel 

 

[dirty1_1garry]

Xformer dimensions:

 

[dirty1_1garry]

And here it is - I unsoldered xformer wires from the circuit and measured resistance between primaries and secondary.
I checked resistance with two different multimeters, both gave same results.

 

[dirty1_1garry]

Done some audio samples of 3 different DI's on bass gtr.

- Passive Jensen DBE-1 DI with neve preamp.
- R***i tube DI
- Custom tube preamp with Di inputs

JB bass 48khz/24bit:
http://yadi.sk/d/unksk_oK4L-zO

My opinion:
R***i - nice warm character, but not so transparent as custom pre.
jensen DI/neve - not so transparent as other two, slightly soapy sound.
Custom - most transparent and clean but without strong character.

What's your opinion?! 

 

[CJ]

you are the Man!

thanks again, Harry,  ;D

that is indeed the 100 EI lam, sq stack,

big enough to be a OPT for a pair of EL 84 tubes, sheesh,

average turn for 100 EI is 14.5 cm = 5.7 inches,

we are running 16 ma plate current,

#38 good for 15.7 ma,

lets bump up to #36 which handles 25 ma just to be safe and calculate turns based on DCR, turn length, and ohms per 1000 ft for #36,

ohms per 1000 ft of #36 is 415 at sea level, 

957 ohms pri / 415 ohms per 1000 ft = 2.3,  2.3 *1000 = 2,300 feet of #36,

convert to inches and divide by turn length, 2,300 * 12 inches = 27,600 inches

27,600/5.7 = 4,842 turns,

winding area for a 100 EI bobbin is  1 1/4" by 3/8"  = .4687 sq in.
copper area is split between pri and sec, and we need room for tape since this thing does 60 k hz which translates to about 7 sections of coil, 

0.492/2= 0.246 sq inches, subtract for tape and call it 0.2 sq inches for pri  wire,

#36 = 31,000 turns per sq inch, so we have room for .2 *31,000 = 6,200 turns,

we guessed 4,842 turns based on DCR and 6,200 turns based on wire size,

let's call it 5000 turns and see what the Henries are for a gapped 100 EI core,

might be 4000 turns looking at how full the bobbin is,

for a 10 mil gap we get 37 Henries, for a 5 mil gap we have 50 Henries pri L,

50 Henries @ 20 Hz = 6.28 * 20 * 50 = 6 K pri Z, if 10 mils then 4.7 K,

if turns ratio is 3:1, then Z ratio is 9:1,

6K/9 = 667,  4.7/9 = 522 ohms, seems ballpark to me,

gauss is 500 AC and 1000 DC, total flux is 1500 gauss so we are loafing on a steel core,

sec turns will be 5000/3 = 1667,

1667 * 5.7"  = 9500 inches  9500/12= 792 feet of wire,

what size?

132 ohms, 792/1000 = .792,  try #32, ohms per 1000 = 163,

0.792 * 163 = 129 ohms, Bingo!  so 5000 T #36 and 1667 T #32 on a 100 EI,

Garry, if you have time, measure AC current at 100 hz, this will tell us inductance,
then we can see how close we are to guessing turns,

should be about 13 VAC/6.28*100(f)*50 (Henries)
=
13/628*50=13VAC/31,400 Z-ohms = .000414 = 414 uA

thanks!  test procedure below>

 

[dirty1_1garry]

Garry, if you have time, measure AC current at 100 hz, this will tell us inductance,
then we can see how close we are to guessing turns,
should be about 13 VAC/5000Z= 0.0026 = 2.6 ma

Just wonna clarify test procedure. is it correct?!

 

[dandeurloo]

You guys are blowing my mind.

 

[CJ]

yes Garry, your circuit is correct,

may as well get the DC current also, just flip the meter switch to ma-DC,
but AC is more important, we can get Henries from
the AC current,

if you want, use 20 hz and you will get a higher current reading, like 13/6.28*20f*50H=

13/125.6*50 = 13/6280= 2.07 ma,

for extra bonus points, sweep the thing from 500 to 1500 hz and find the freq with the lowest current draw, this will tell us how big the gap is, should be around 1 k hz,

this will be the peak inductance point,

bigger gap moves the Q peak up in frequency, 5 mil gap might peak at 900 hz, 10 mil gap might peak at 1 k hz,

current should be very low, in the micro-amps range, 10 uA +/- 5 uA,

thanks!

 

[dirty1_1garry]

Did IAC test.
But did it before read CJ last comment so did not measured IDC and sweep frequencies.

 

[dirty1_1garry]

and here is a pwr scheme!!

 

[CJ]

Cool Garry!

DC on the heaters and mega filtering on the B+, i bet it is quiet,

ok we have 227 Henries @ 100 Hz and 22 Henries @ 1 k hz,

this means that there is a very small gap as we have large inductance down low and the DC is taking over quickly and rolling off the high end,

looking at the shot of the xfmr, i see little or no gap, the butt stack will have about .005/6(MPL for 100 EI) = 0.0008 natural air gap, add maybe 1 or 2 mils of KP and we get our 227 Henries @ 100 Hz,

no 100 EI in the house yet, so will use 1 1/8" stack of 75 EI to simulate the core, wind 5000 T and check current vs freq,

thanks again Garry! 

 

[CJ]

ok we have a 5000 turn coil wound on a 75EI bobbin, .75" by 1.25" to simulate a sq stack of 100 EI,

maybe this core will sound better, who kows,

we need some DC to check incremental perm/inductance, but if you use a 300 Volt DC  supply thru, say, a 18K resistor, then the AC signal being injected into the primary will go thru the resistor and get swamped out by the 25uf/400 volt filter cap,

so to really get a good graph of what the inductor/xfmr will be doing in the circuit, you have to wire up the actual circuit, inject AC into the grid, and measure current vs volts across the inductor at all the freqs,

i have used a HV supply, it will give you some idea of what the coil/core is doing, but when compared with an actual in circuit test, you will see large differences when the inductor reaches maximum impedance, if you have a coil that will do 2 meg-ohms at 10,000 hz, then that 18 k resistor really becomes a problem as far as getting an accurate plot,

here is a 6DJ8 already wired and ready to go,

 

[CJ]

ok playing around with this 6DJ8,

be careful if using this tube, i did not drop my pwr supply volts down, so i had 275 on the plate of the DJ8, it started turning weird colors of blue and green, not like a 12AX7a which will take a lot more juice than advertised, this DJ8 is sensitive to anything over 170 volts on the plate,

tweaked the cathode resistor for 16 ma, ended up with a single 220 ohmer with both cathodes tied together, i tried splitting the cathodes but noticed no increase in stability,

now i am wondering if the schematic needs correcting, that 6N1P is shown with separate 220 ohm cathode resistors, but since the DJ8 cathode resistor turned out to be 220 ohms for both, and the REDD 47 has a 220 ohm resistor for both cathodes also, maybe the schematic should have both cathodes tied together, always wondered why they went through the trouble to use add 2 more parts with no apparent reason,

maybe the 6N1P responds differently to cathode resistors, somebody with a 6N1P could maybe verify this,

ok, circuit running, plate volts dropped down to 166, remember that the transformer DCR is much lower than a plate resistor, so if you apply 300 to the xfmr, the plate will get 284 volts compared to maybe 120 with a 150K plate resistor,

#36 wire seems to be the ticket for the pri, tried #38 but DCR was 1300 ohms for 5000 T, tried # 34 but it took up too much room,

going to run some AC thru this and see what the inductor current is, no sec yet, just a pri on the 75 EI stack, so we will have to wind a sec on there to hear the results, but first going to play with the air gap and tweak the turns to get 91 uA at 100 hz,

 

[CJ]

alright, we get 109 uA @ 1000 hz with 13 VAC across the 5000 T pri coil,

so we have about 15% more AC current, which means we need a bit more inductance,

could be the different core since we are using 75 EI instead of 100 EI, even though we have about the same inductance per turns with the bigger stack height that we are using, according the the mag metals coef. for Henries,

so if we want to use the 75EI, we just add about 200 turns, remember that Henries go up as a square of the turns, so we do not have to add that much copper to get where we want to go,

so 5200 turns #36 for the pri seems good,

i did not get the same numbers for 100 hz as Garry measured, he got 94 uA, i got about 350 uA, so wondering about that, most coils that i have checked do drag more current at 100 hz as opposed to 1000, so i am going with the 91 uA @ 1000 hz as the correct reading,

turns ratio will go up to 3.12 if we keep sec turns at 1667, this means a better load for the tube, so lets try 5200:1667 for the prototype,

with 5000 T and 16ma, we have a serious electro magnet here, about 12 to 14 Oersteds, (H-Oersteds = 1.777NI/Magnetic Path Length)so we do not need to build an over sized channel frame to hold the lams, the DC sucks them right down on the core,

time to wind the xfmr and clean up the chassis,

only had to inject 350 milli-volts instead of .775 to get 13 volts across the pri,

might have something to do with the 6DJ8 vs the tube from the East,

we have a hefty core so only a 1 mil gap will be fine, or even no gap at all as long as we butt stack this thing,

going with a 10K audio for the vol control, 1 k seems a bit harsh, we can always crank down to 1 k if there is some sort of sonic improvement,

lifting up the xfmr by the I bars to illustrate the electro-magnet effect,>

 

[dirty1_1garry]

ok we have 227 Henries @ 100 Hz and 22 Henries @ 1 k hz,

IAC on primary seems pretty close on 100Hz and 1kHz frequencies. Why inductance so difference? I'm not a hight quality transformer specialist, so here is a question))

CJ, you''ve made ​​a prototype already!  Great result. 

Could you tell us a transformer parameters for the final?

There are some crazy guys in my city that could try to made that transformer too. Or may be it's possible to find appropriate parameters in sower/cinemag/lundal/edcor/hammond assortment.

 

[abbey road d enfer]

only had to inject 350 milli-volts instead of .775 to get 13 volts across the pri,...
...going with a 10K audio for the vol control, 1 k seems a bit harsh, we can always crank down to 1 k if there is some sort of sonic improvement,

That may be the reason of your different results. The load wil reflect in the primary. I guess the idea in the original design is making the output level not so much dependant on what it is connected to. With a 10k pot, the output level may vary as much as 20dB between a hi-Z line input and a Neve-style mic input.
At any frequency above about 50Hz, the reflected load is dominant over the inductance.