awesome Doug! that is a double stack of 625 EI in that can,
Langevin 5116 / Collins 356A Output Transformer Tertiary Feedback?
- Thread starter Braden Sauder
- Start date
Help Support GroupDIY Audio Forum:
Damn, I love this place!
practice coil: assume 625 EI twice stack, bobbin has 1/4" flange, we can use about 0.1" of that for wire if equal current density between pri and sec, we need to use some of that space for insulation and other losses, so we can not simply divide that flange dim in half,
old layered paper like they use in the original has margins so we use 3/4" winding length per layer,
if we compute square feet of pri wire we can use some nifty charts in RDH4 (page 1412) to play around with turns(AC and DC flux and Henries) and DCR,
.1" tall by .75" wide = .075" sq ft of wire, RDH4 says we can spin 62,600 turns of #39 wire (ball park for 6 ma DC) per sq inch, so 0.075" x 62600 = 4695 turns, this would yield 5 K Gauss AC and 10 K Gauss DC (lots of amp-turns) Henries would be 40 H for a core perm of 300 (normal for a gapped core)
to get DCR we need cubic inches, just multiply sq inches by 1 average turn length, about 4.25" for a double stack, so .075" x 4.25" = .318 cubic inches, now we get DCR from same chart in RDH4 page 1412,
#39 says 4332 ohms per cubic feet, so .318 x 4332 = 1377, too low for what Doug reads off his coils,
try #40, still reasonable for 6 ma, #40 has 6770 ohms per cube, so .318 x 6770 = 2152 ohms, now we are getting somewhere, what if we bring the turns down to get 2000 ohms? 2000/2152= 93 percent,
.93 x 4695 T = 4363 T. since we are using smaller 40 wire, we will have room for the FB wind and the insulation it needs on both sides.
now we go back and plug those turns into the calculator to get flux and Henries,
looks like 5.7 K Gauss at 100 V-rms and 20 Hz, reasonable.
DC flux is 9.4 K Gauss. It is normal to have more DC flux than AC flux in coils of this type.
Total flux is then 5.7 + 9.4 = 15.1 KG at peak input, so we are close to saturation, (18 KG) but still have 3 KG of headroom.
Henries for 4363 T on 1.25" of 625 EI (5/8" tongue) will be about 40 H (perm = 350 for gapped core)
we had a figure of 35 H mentioned, but that was at 120 Hz, a gapped coil will ramp up a bit in terms of Henries as frequency decreases but not as steep as a non gapped core so 40 H is reasonable.
Now we can go back and get sec turns, 4363 T/ 5.25 = 831 T for the 600 Ohm wind, 150 ohm tap will be 831/2 = 416 T
F/B will be .12 x 831 = 99 turns. lets adjust some numbers, round up the F/B to 100 T, 150/600 = 416/832, so pri = 4368 T.
now we go back and use sec DCR numbers posted to get wire size, 70 Ohms for 832 T,
MLT = 4.25" x 832 = 294 ft., how many ohms if we had 1000 ft of wire?
1000/294 = 3.4 x 70 = 238 Ohms/1K ft. that's probably #34,
help here: http://www.coonerwire.com/magnet-wire/
FB = 100 T x 4.25 = 35 Ft. 1000/35 = 28.57 x 4 ohms = 114 Ohms/1000 ft , probably #31
summary:
Pri: 4368 T #40 AWG
Sec: 416/832 T #34 AWG
F/B: 100 T #31 AWG
Core: 1.25" of 625 EI Butt Stack , .002 gap, 29gaM6 or Cold Rolled.
Structure: Try Sec-Pri-FB-Pri
you could also use a 1" stack of 87 EI (7/8" Tongue) and get about the same flux density.
.625" x 1.25" bobbins for 625 EI are hard to source, 1 " bobbin for 87 EI available at Edcor.
old layered paper like they use in the original has margins so we use 3/4" winding length per layer,
if we compute square feet of pri wire we can use some nifty charts in RDH4 (page 1412) to play around with turns(AC and DC flux and Henries) and DCR,
.1" tall by .75" wide = .075" sq ft of wire, RDH4 says we can spin 62,600 turns of #39 wire (ball park for 6 ma DC) per sq inch, so 0.075" x 62600 = 4695 turns, this would yield 5 K Gauss AC and 10 K Gauss DC (lots of amp-turns) Henries would be 40 H for a core perm of 300 (normal for a gapped core)
to get DCR we need cubic inches, just multiply sq inches by 1 average turn length, about 4.25" for a double stack, so .075" x 4.25" = .318 cubic inches, now we get DCR from same chart in RDH4 page 1412,
#39 says 4332 ohms per cubic feet, so .318 x 4332 = 1377, too low for what Doug reads off his coils,
try #40, still reasonable for 6 ma, #40 has 6770 ohms per cube, so .318 x 6770 = 2152 ohms, now we are getting somewhere, what if we bring the turns down to get 2000 ohms? 2000/2152= 93 percent,
.93 x 4695 T = 4363 T. since we are using smaller 40 wire, we will have room for the FB wind and the insulation it needs on both sides.
now we go back and plug those turns into the calculator to get flux and Henries,
looks like 5.7 K Gauss at 100 V-rms and 20 Hz, reasonable.
DC flux is 9.4 K Gauss. It is normal to have more DC flux than AC flux in coils of this type.
Total flux is then 5.7 + 9.4 = 15.1 KG at peak input, so we are close to saturation, (18 KG) but still have 3 KG of headroom.
Henries for 4363 T on 1.25" of 625 EI (5/8" tongue) will be about 40 H (perm = 350 for gapped core)
we had a figure of 35 H mentioned, but that was at 120 Hz, a gapped coil will ramp up a bit in terms of Henries as frequency decreases but not as steep as a non gapped core so 40 H is reasonable.
Now we can go back and get sec turns, 4363 T/ 5.25 = 831 T for the 600 Ohm wind, 150 ohm tap will be 831/2 = 416 T
F/B will be .12 x 831 = 99 turns. lets adjust some numbers, round up the F/B to 100 T, 150/600 = 416/832, so pri = 4368 T.
now we go back and use sec DCR numbers posted to get wire size, 70 Ohms for 832 T,
MLT = 4.25" x 832 = 294 ft., how many ohms if we had 1000 ft of wire?
1000/294 = 3.4 x 70 = 238 Ohms/1K ft. that's probably #34,
help here: http://www.coonerwire.com/magnet-wire/
FB = 100 T x 4.25 = 35 Ft. 1000/35 = 28.57 x 4 ohms = 114 Ohms/1000 ft , probably #31
summary:
Pri: 4368 T #40 AWG
Sec: 416/832 T #34 AWG
F/B: 100 T #31 AWG
Core: 1.25" of 625 EI Butt Stack , .002 gap, 29gaM6 or Cold Rolled.
Structure: Try Sec-Pri-FB-Pri
you could also use a 1" stack of 87 EI (7/8" Tongue) and get about the same flux density.
.625" x 1.25" bobbins for 625 EI are hard to source, 1 " bobbin for 87 EI available at Edcor.
It occurs to me that the tertiary will be around 5 Ohms nominal. Not that you would load it in 5 Ohms (it would absorb all the amp's power) but you could load it in 50-100 Ohms.
Instead double the estimated turns. 20 Ohms. Load it in a 200 Ohm pot. 200r is "low" compared to first stage cathode impedance. Take NFB from the pot wiper. Gain can be adjusted from half the original design toward infinity (actually to the open-loop no-NFB gain). You could dial-in your 40dB without tedious re-re-winding.
Instead double the estimated turns. 20 Ohms. Load it in a 200 Ohm pot. 200r is "low" compared to first stage cathode impedance. Take NFB from the pot wiper. Gain can be adjusted from half the original design toward infinity (actually to the open-loop no-NFB gain). You could dial-in your 40dB without tedious re-re-winding.
5% makes more sense,
that would put the wire size for the FB wind at #34, same as the sec wire, easier to use two gauges instead of three, so a revised turns count for 5% FB would be 42 turns of #34.
Is the p/n on that output 313A? if so it is 16 K, not 16.5 K
here is a link to a parts page:
http://www.technicalaudio.com/pdf/Langevin/Langevin_Catalog_1952_product_listing.pdf
16 K would mean 26.667 Z ratio = 5.16:1 turns ratio, so Sec and FB turns would be 423/846 and 42 .
Pri: 4368 T #40 AWG
Sec: 423/846 T #34 AWG
F/B: 42 T #34 AWG
Core: 1.25" of 625 EI Butt Stack , .002 gap, 29gaM6 or Cold Rolled.
Structure: Try Sec-Pri-FB-Pri
you could also use a 1" stack of 87 EI (7/8" Tongue) and get about the same flux density.
that would put the wire size for the FB wind at #34, same as the sec wire, easier to use two gauges instead of three, so a revised turns count for 5% FB would be 42 turns of #34.
Is the p/n on that output 313A? if so it is 16 K, not 16.5 K
here is a link to a parts page:
http://www.technicalaudio.com/pdf/Langevin/Langevin_Catalog_1952_product_listing.pdf
16 K would mean 26.667 Z ratio = 5.16:1 turns ratio, so Sec and FB turns would be 423/846 and 42 .
Pri: 4368 T #40 AWG
Sec: 423/846 T #34 AWG
F/B: 42 T #34 AWG
Core: 1.25" of 625 EI Butt Stack , .002 gap, 29gaM6 or Cold Rolled.
Structure: Try Sec-Pri-FB-Pri
you could also use a 1" stack of 87 EI (7/8" Tongue) and get about the same flux density.
Been going back and forth with David at Cinemag about this. He's been focusing on seeing what it would take to duplicate the design, rather than pointing us to a sub-optimal use of the 9600, so I sent him the data/calculations CJ came up with and the measurements/thoughts from Doug and PRR. He has a "confidential source" for some of the Langevin engineering information. He actually sent me the engineering drawing, but it lacks key information he needs. (I'm hesitant to post it since he called it "confidential"). Here are the last two emails I received from him on it (minus the engineering drawing, which I'm happy to PM to CJ/EMRR/PRR if desired). This was in response to the figures CJ calculated.
Hi Bryan;
625EI bobbins up to 1" are available from multiple sources. I doubt that a larger stack is necessary anyway. This coils structure seems to be too simple for a high quality amplifier. I did not see any tear-down information on GroupDIY, which can be a good resource for such things since we want to come close to what it originally was. Show me were you got your information. I will do some digging with a confidential resource that I have access to, in case Langevin actually specified it in detail or made it themselves in house.
All the best,
David
And the email that accompanied the drawing, after I sent him the actual thread:
The detailed transformers winding specifications have not been cataloged yet, if they exist anymore. The input transformer was 1:10.8, but the specification sheet does not say if that was with the primary tied for 600 Ohms or 50, or even 150. The maximum signal level was only -10dBm, which might indicate that the turns ratio was for the 600 Ohm configuration given that it looks like it used 375EI laminations. We need more information.
David
Not sure why he talks about the input transformer above. When I asked he didn't directly answer. I'm thinking it's a typo, but not sure. When I asked what he needs at this point, he said either a sample unit to do some testing/measuring on, or a tear down. I suggested he consider checking into the Collins version with its ADC built output, but he hasn't responded since last night. Guess we'll see where it goes and I'll report back.
BT
Hi Bryan;
625EI bobbins up to 1" are available from multiple sources. I doubt that a larger stack is necessary anyway. This coils structure seems to be too simple for a high quality amplifier. I did not see any tear-down information on GroupDIY, which can be a good resource for such things since we want to come close to what it originally was. Show me were you got your information. I will do some digging with a confidential resource that I have access to, in case Langevin actually specified it in detail or made it themselves in house.
All the best,
David
And the email that accompanied the drawing, after I sent him the actual thread:
The detailed transformers winding specifications have not been cataloged yet, if they exist anymore. The input transformer was 1:10.8, but the specification sheet does not say if that was with the primary tied for 600 Ohms or 50, or even 150. The maximum signal level was only -10dBm, which might indicate that the turns ratio was for the 600 Ohm configuration given that it looks like it used 375EI laminations. We need more information.
David
Not sure why he talks about the input transformer above. When I asked he didn't directly answer. I'm thinking it's a typo, but not sure. When I asked what he needs at this point, he said either a sample unit to do some testing/measuring on, or a tear down. I suggested he consider checking into the Collins version with its ADC built output, but he hasn't responded since last night. Guess we'll see where it goes and I'll report back.
BT
Remember as I've said elsewhere, the timeline of documents suggests the GE version may have been the first, and it has a 25K output primary. The earliest lineage is the Langevin 116, though many overall values are different, any of the outputs would work in it too.
Max out of +18dBm is one parameter probably not discussed in this thread.
Max out of +18dBm is one parameter probably not discussed in this thread.
Okay, good info Doug. Will pass that on.
bluebird
Well-known member
I can wind this. I will have to get the wire but I have the (625) lambs and the bobbins and tape for the layer winding. I do have 41awg 39awg 28awg.
CJ if you want to give me the turns with those wire gauges I could do one sooner than later. My tape is .005 in.
I can send the transformer to rackmonkey to try in the circuit. (or Doug if you want to try it one of those original units)
CJ if you want to give me the turns with those wire gauges I could do one sooner than later. My tape is .005 in.
I can send the transformer to rackmonkey to try in the circuit. (or Doug if you want to try it one of those original units)
Thanks for offering that up, Bluebird. I'm not the OP, but would be willing to build the circuit for a test as it's something I've been interested in for a while. I'm in no hurry on that though so maybe let's see where David goes with this first. Given the general interest in this circuit, it would be a good thing if Cinemag could come up with something.
bluebird
Well-known member
Yea, I could just wind one up for testing on my home brew machine. Not into making them for sale, too much work! So perhaps the Cinemag route is better. If you do get David to make a version I would be interested in the sound difference of the two winding styles. I can make one with tape layers between each winding which is probably how the original was made. I don't think this would be practical for Cinemag to do it that way. Cinemag can actually achieve better specs with the computer winders they have without the layers. But I am curious about the difference in sound.
Because in order to analyse the gain structure, it is necessary to know the ratio of both input and output xfmrs.rackmonkey said:
The American standard appears to have been quoting gain at the highest input and output impedances. I get this from measuring the gain of many.
Here's another level measurement I found on a raw transformer:
16K5 14.1 dBm
600 0 dBm
tert -19.4 dBm
Obviously not dBm, but that's what the meter is labeled as. Interpret at will.
Here's another level measurement I found on a raw transformer:
16K5 14.1 dBm
600 0 dBm
tert -19.4 dBm
Obviously not dBm, but that's what the meter is labeled as. Interpret at will.
db calculator says 10 volts on the 600 wind and 1.07 volts on the FB wind = -19.4 db
so it looks like a 10:1 turns ratio on the fb, 10^2 = 100 Z ratio so fb ohms are 600/100 = 6 Ohms, close to the PRR speculation of 5 Ohms,
so it looks like a 10:1 turns ratio on the fb, 10^2 = 100 Z ratio so fb ohms are 600/100 = 6 Ohms, close to the PRR speculation of 5 Ohms,
cool on that looks like we have the fb well into the ball park,
18 db, that would mean about 40v / 8 v voltage ratio pri to sec,
100 v / 20 v is 26 db input level, so they were putting the Gauss down into a more reasonable level,
and probably a smaller core but Henries would drop which means more pri turns required, DCR would probably stay about the same as the mean length turn of wire would go down .
So maybe an EI stack of 1" is what is in there,
Doug, have you done a freq sweep on that OPT?
18 db, that would mean about 40v / 8 v voltage ratio pri to sec,
100 v / 20 v is 26 db input level, so they were putting the Gauss down into a more reasonable level,
and probably a smaller core but Henries would drop which means more pri turns required, DCR would probably stay about the same as the mean length turn of wire would go down .
So maybe an EI stack of 1" is what is in there,
Doug, have you done a freq sweep on that OPT?
Similar threads
