altec 9475a clone

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

rafafredd

Well-known member
Joined
Jun 3, 2004
Messages
2,409
Location
Rio, Brazil
So there can never be clones enought.

And this one is from an old design, so it´s VERY OK to clone.

I´ve got some UTC A22, transistor interstage or output, 500/500,125 that can take up to 20maDC on the windings, so I think it´s a good candidate for the output.

let make some questions:

1 - I would like to know opinions about transistors substitutions. I will try to get the 2n3900a but the AL ones seems to be ALTEC part numbers, so if I could use something like BD 135,139, 237 or any other common parts at the output it would be nice!

2 - About the input transformer, what do you think would be a nice turns ratio? Maybe 1:4? Or should I go higher?

3 - Also, should I include the .047uF C2 on the secondary of the transformer or is it just used with that particular transformer?

4 - I also hope that if I match R1 and R2 close, there will not be any current throught the secondary of the input transformer, am I right?

5 - should 1N4148 work for CR1 and CR2?

6 - Any other things I did not considered?

And now for the schem:

http://hps.infolink.com.br/rafafredd/altec9475asch.jpg

Thanks guys!
 
Sorry for the bump, but I hope to get at least one input...

I know, I did lot of questions, again... :?

Forget question five. I think I´ll put some 1N400x there.
 
The input transformer must be balanced push-pull secondary. Most current production assumes that one end of the secondary is grounded, and puts the shielding closer to that side. Using that near-ground end of the secondary as a "live" wire won't much affect the frequency response, but will raise 2nd harmonic distortion at high frequencies. Which may be a nice thing, but isn't what this preamp was made to do. So where are you going to source an input transformer? This beast is too complicated to build unless you want an EXACT duplicate of the original. There are cheaper ways to get rising even-order distortion.

Transformers like that were used on the early Langevin transistor amps. But do you want to cut-up a healthy Langevin to build what looks like a mediocre Altec?

The cap on the input secondary center-tap is odd. It isn't to load the leakage inductance, like the Zobel network often used on hi-Z windings. It appears to force push-pull balance at high frequencies while allowing the winding to float at low frequencies and at DC. That gets into the wacky DC biasing, which at a glance won't work unless each pair is pulled from the same lot and very tightly matched (and I don't think beta-matching is what they want; it may just be all they had).

In theory there is no DC on the output windings. In fact I suspect an unmatched pair will give substantial DC current, possibly 50mA. And beta-selection, as called for in the notes, isn't sure to avoid that.

Forward gain without the transformers is about 1:27, or 29dB. And I suspect the output tranny is wired 1:1:1:1, so on 600 ohm load it runs unity gain. Such preamps had total gain of 35dB to 45dB. So the input tranny must be 6dB to 16dB gain, or 1:2 to 1:6 turns ratio, just to meet the usual gain spec.

Noise matching was primitive in those days. Often a little futile because the transistors were so dirty and erratic. With modern eyes assuming modern silicon, the first-stage collector current suggests a source impedance of 5K-20K per transistor, or 20K-80K total. That suggests at least a 1:10 input transformer. That also leads to 49dB overall gain. The output stage will clip at +25dBm, so +18dBm is a likely output for rated THD. +18dBm minus 49dB gain gives input overload of -31dBm, 22 mV in 600 ohms, about 11mV in 150 ohms. That is a very low input overload level: classic preamps should take 30mV, and in modern work we often feed them hundreds of mV. We could reduce the input transformer ratio, the slight noise mismatch is probably no big deal, but we are veering more and more away from what Altec built.

I'll stare at this some more, for morbid curiosity, but my sense is that it isn't a great preamp, just the best Altec could do in the dim days of the dawn of sand-state. If it is known to be a good preamp, then it is for subtle things not documented on the schematic (exact core and winding of transformers, factory balance or lack thereof) and an exact clone isn't going to be easy. An inexact clone with a different sound is possible, but there are much easier plans that are less likely to smoke on the testbench.

If you really must build this thing: for outputs, I don't see why 2N3055 would not work. They typically have beta over 100 at 100mA, not far from the spec. I'd have to trace the DC bias to see why they are so fussy about beta, and if there is a better way to balance the DC than by unreliable beta-matching.

For flame-test, I sure would throw 10 ohms in the emitter of each output transistor. That slightly changes the biasing of the input transistors, but not enough to hurt, and may prevent runaway unbalance or at least give clear indication of problem (smoked resistor).

Oh, both diodes should be 1N1001-series. They carry 100mA which is at the limit of small diodes. Otherwise nothing fussy, and 1N1001-1N1007 diodes are cheapest.
 
Thanks for all the comments PRR. Much aprecciated!

I was thinking about using any input transformer with two secondaries. If I wire it the right way, doesn´t it end working as a balanced winding?

But after all you said, I don´t think it would be a good idea to go far with this project. This preamp is know to be the Altec response to Langevin´s AM16, but maybe they are not all this marvelous. I should buy the sowter trannies for the AM16 and try my clone of the Langevin thing instead.

I´m sure I will find another good use for my six UTC A22´s. It´s just that I have been looking for this Altec schem for a long time, thinking about a clone. But in the end, it doesn´t seems to be a good choice for a clone anyway, unless I can put my hands on a original input transformer someday.

It´s always nice to hear what you have to say about a circuit anyway!

Thanks!
 
> any input transformer with two secondaries. If I wire it the right way, doesn´t it end working as a balanced winding?

For low frequencies, yes. For high frequencies, the stray capacitances have to be balanced. Jakob may be able to explain better, and know if readily available trannies are sufficiently capacitively balanced.

> Altec response to Langevin´s AM16,

I was wondering at the superficial similarity.

But the AM16 has more complicated output transformer for Macintosh-like output stage linearity. And I think it has one more stage of transistors so it does not try to scrape every last bit of gain from each transistor.

And the real difference is: the AM16's DC bias controls the side-to-side DC current balance. This Altec is clearly two single-ended amps (the classic 2-transistor phono preamp with Darlington second stage) just stuck together. In single-ended form the 2-tranny phono preamp has very good DC bias control (at least for temperatures people can stand to work in). But when you stick two of them together, it is like putting two horses on one cart with a swivel cross-bar: the horses will never be perfectly matched and one will be in front of the other. It looks like Altec first tried hand-selecting pairs of transistors to absurdly tight specs, but beta is very variable with temperature and even age so this didn't work. Oh, and hand-matched the critical overall feedback resistors because they throw DC error into the input stage. Then they added all those 1Meg feedback resistors, and apparently got it "good enough". But what these really do is feed back the Vbe of the output transistors. Which is a very poor indicator of output stage current. Especially with "big" transistors that have low parasitic resistance. And misleading if the two outputs are not matched for Vbe (different working die sizes, doping resistivity).

Even for 1965, this is clumsy design for production. Not unusually clumsy, and it looks worse through 40 years of hindsight than it would have in 1965. But I don't like it. A circuit should "work" with about any appropriate-size/beta transistors you put in it. I suspect that a random pick of 1965 transistors here could give one side working at 20mA and the other side at 80mA, nearly single-ended operation for push-pull price.

If you breadboard it, do start with 5 ohm or 10 ohm emitter resistors in the output stage. In fact the original transistors may have had several ohms parasitic emitter resistance, a parameter that will typically be much lower and not clearly specified on modern transistors. May as well put in an actual resistor. The loss of gain is small or none. (Emitter resistance will reduce output stage gain, but because this circuit is starved for current gain and heavily loading the first stage, the resistors will actually increase first-stage voltage gain.)

Correction: maximum output at clipping may be +31dBm. So it may be low distortion to +24dBm, and 6dB above that with no gross clipping. If the input transformer is 1:4, for 40dB overall gain, the max input level is 150mV, a nice high value.

Some of the Langevin AM1x modules were really 10-watt outputs with an option to cut them down to 1 watt. That's very different from what is really a couple of 1/4-watt circuits slapped together in loose push-pull.
 
If you're looking to use those A22s, I have some Dynair line amps that use a simple, four-transistor push-pull circuit that looks like a solid-state version of a Langevin 5116B. It uses four 2N697 transistors, which were still readily available last time I checked, and it uses an A22 for the output. This circuit could be worth considering.

The catch is, I couldn't get a schematic (not even from the founder of Dynair, who doesn't have it) and I haven't found the time to finish tracing it out and drawing my own schematic yet.
 
With apologies to John Hall's memory, some facts gleaned from (recent) personal experience and Altec and Langevin manuals and catalogs in my possession:

John Hall claims to have designed both the Langevin and the Altec preamplifiers in question. 

I've dealt with over 40 of these modules in the last few years with only one transistor and several of the power supply electrolytics ever being bad.  The current draw is almost always spot-on with the manual specifications, which should refute any notions of poorly matched or selected parts.  I'm not sure how one could call this clumsy design, being well-established methodology in tube design and perfectly applicable to SS.

The 470A / 475A / 9470A / 9475A are all the same amp with only the slightest circuit variations, those being a change in some shield connections, an overall polarity inversion, and a switch from matched pair single case 2N2716 transistor pairs to individual 2N3900A's. 

NOTE THERE IS A LATER SQUARE 470A THAT IS NOT THE SAME THING!  APPARENTLY ALTEC STARTED USING THAT DESIGNATION FOR A TELEPHONE MODULE AFTER THE ORIGINAL 470A WAS CHANGED TO 9470A.  LOOK UP PICTURES IF IN DOUBT!  A 470A SHOULD LOOK EXACTLY LIKE A 9470A IF IT'S REALLY THE SAME THING!

The 470A / 9470A are meant for rackmounting, the 475A / 9475A were meant for retrofitting 250SU consoles when you were ready to throw out those horrifyingly bad :) 458A and 459A tube modules.  They were also stock in the SS 250T3 consoles.  There are less external connections on the 9475A's, but everything is there on the circuit board.

Judging by the catalogs and schematic notations, the Langevin came on the market in 63, and the Altec drawings are dated 9/64.  Both types appear to have been marketed continuously until at least 1972.
The Langevin AM-16 is the only other American made preamp I know of using a push-pull class A discrete transistor circuit. John Hall recently stated "It (Altec 9475A) is an updated Langevin AM-16. Has wider freq. resp and less distortion.  The input xfmr was much better than the AM-16."  They are the only two stock manufactured circuits with any similarity.  Every other manufacturer went with single-ended designs.  The Altecs and Langevin both have nearly identical power requirements, harmonic distortion ratings and output power ratings (+27 dbm).  Both have -127 dbm noise ratings.  -22 dbm is the input overload level for the Altec, which matches pretty much everything else manufactured at the time.  The AM-16 has about 7db less gain than the 9475A, as I'll mention again below.  This gain difference relates to a measured lower input headroom for the AM-16.  Quoted gains are 45 db (input terminated) for the Altec and 45 db (termination unspecified) for the Langevin.  My measurements show 51 db for the Altec when wired 150 in and 600 out with no input termination.  The AM-16 must have a higher amount of negative feedback and/or lower stage gain due to the use of 8 transistors rather than 6 as found in the Altecs.  Gain specs and basic theory are pretty much the same for the input iron (2db more gain in the Langevin input at 19 db), but the outputs are totally different in theory with the Altec being a straight 1:1 repeat coil and the Langevin being 1:1:2 with McIntosh style output primary coupling which makes for very different feedback paths.  The audio path crossover in the middle of the AM-16 circuit should contribute to a different push-pull / feedback harmonic distortion cancellation profile, and the additional transistor stage should make a pretty large difference to the AM-16 sound also, but not as much as the iron.  The Langevin circuit seems more overkill to me, though in elegant ways.  It adds a pot for adjusting harmonic distortion; this is good in theory, but what if it's out of adjustment?  The freq. (and implied phase) charts as published by the companies are certainly night and day differences.  The AM-16 claims to be down 6db at 40kHz, where the Altec's -3db points are listed as 2.5 Hz and 100kHz.

The Altec circuit would be very easy to build as there is nothing unusual about the transformers. The Langevin input is not unique, but it's output is specialized.  The Altec transformers do indeed have Peerless stamps and part #s, and match the profiles of other Peerless products to a very high degree.  I see no reason for Peerless (one of the best in the business) to farm that winding work out as John Hall has stated.  The Altec input in particular is practically identical to the 458A / 459A tube input and the previous generation 4629 input, at least to the point that one could replace either earlier type with the 9475A type.

As to sound, both are great preamps, but to my ears the 9475A is the more hi-fi unit with an overall faster sound on transients, tighter bass and better top and bottom octave extension.  The AM-16 in comparison has a slightly slower, milky/gauzy sound with a slightly more rounded tone; more in the direction of the tube gear that both replaced.  The AM-16 has a slight upper mid/treble bump while the 9475A is more purposefully flat.  These comparisons may or may not seem hair-splitting to the individual end user.  Many people really love the slightly softer more mid-range focused sound of the AM-16.  The AM-16 seems to win the midrange competition over the 9475A.  The AM-16 had about 7 db less total gain in a direct comparison under identical and matched standards, including same power supply at same time.   

The only remaining piece of the puzzle would be the question of how the AM-16 came to be so well known, and the Altec did not.  I suspect it relates to fate regarding placement of AM-16's in certain hit producing studios where word would have spread.  I would think more Altecs ended up in radio, TV, and film studios due to marketing approach differences between the companies. I would suspect the # of Altecs to be higher than Langevins.

I think I'm working on a book, so compiled contents copywrite 2005 Doug Williams
 
Doug In reference to the AM16 gain of 45db , it depends on which AM16 you have or if its has been changed .

I have a book from the Langevin factory showing all manufacturing info on the AM16C AM16R AM16S AM16X
The AM16R has 50 db of gain ( as do all my AM16's) the AM16X was ,made for Raytheon and had a carbon mic input!
Only the R and X state in the book what the difference is from a AM16..
Doug contact me if you are interested in more info on the AM16.

Port






-------------------------------------------------------------------------------------




Judging by the catalogs and schematic notations, the Langevin came on the market in 63, and the Altec drawings are dated 9/64. Both types appear to have been marketed continuously until at least 1972.
The Langevin AM-16 is the only other American made preamp I know of using a push-pull class A discrete transistor circuit. John Hall recently stated ?It (Altec 9475A) is an updated Langevin AM-16. Has wider freq. resp and less distortion. The input xfmr was much better than the AM-16.? They are the only two stock manufactured circuits with any similarity. Every other manufacturer went with single-ended designs. The Altecs and Langevin both have nearly identical power requirements, harmonic distortion ratings and output power ratings (+27 dbm). Both have ?127 dbm noise ratings. ?22 dbm is the input overload level for the Altec, which matches pretty much everything else manufactured at the time. The AM-16 has about 7db less gain than the 9475A, as I'll mention again below. This gain difference relates to a measured lower input headroom for the AM-16. Quoted gains are 45 db (input terminated) for the Altec and 45 db (termination unspecified) for the Langevin. My measurements show 51 db for the Altec when wired 150 in and 600 out with no input termination. The AM-16 must have a higher amount of negative feedback and/or lower stage gain due to the use of 8 transistors rather than 6 as found in the Altecs. Gain specs and basic theory are pretty much the same for the input iron (2db more gain in the Langevin input at 19 db), but the outputs are totally different in theory with the Altec being a straight 1:1 repeat coil and the Langevin being 1:1:2 with McIntosh style output primary coupling which makes for very different feedback paths. The audio path crossover in the middle of the AM-16 circuit should contribute to a different push-pull / feedback harmonic distortion cancellation profile, and the additional transistor stage should make a pretty large difference to the AM-16 sound also, but not as much as the iron. The Langevin circuit seems more overkill to me, though in elegant ways. It adds a pot for adjusting harmonic distortion; this is good in theory, but what if it's out of adjustment? The freq. (and implied phase) charts as published by the companies are certainly night and day differences. The AM-16 claims to be down 6db at 40kHz, where the Altec's -3db points are listed as 2.5 Hz and 100kHz.

I think I'm working on a book, so compiled contents copywrite 2005 Doug Williams[/quote]
 
I've put this circuit together to see about making use of some transformers I've had for a few months. Apart from the Altec, a Dynair lineamp seemed to be the only other circuit suited to this transformer. I used parts that I had to get it going, BC550s and 2n2219A for the output stage. Pre-dating this thread John Hall recommended 2n2484 and 2n2219 as replacements. He also mentioned, to put a 910ohm resistor in series with the 9.1K/270pf which increased the total gain by 1dB or less.

On first fire-up there was a severe low end rolloff from 2Khz down, I'm using a Lundahl line input wired 2:1+1. The manual for the 9470a shows a 10k terminating resistor at the input transformer so I did the same which brought back the low end, I'm unsure what makes that neccessary.

The output is wired 500CT:150. With 162R loaded secondary the high frequency suffers, 560R and its fine. The output can't be wired 1:1 anyway but if it was I'm calculating I'd need a 1.1K load on the secondary for equivalent response in the highs.
  Running out at low gain and boosting with the amp 23dB to the soundcard input, the low frequency stands out very nicely...certainly puts some life into an mp3.
  The 500ohm output winding measures 2.4 henries so I was worried that this wasn't comparable to the Altec output iron but thankfully theres no problem in the bass region. The DCR of the same winding is 40ohms so maybe the core is ungapped M6 as the inductance should be far lower for a gapped core of this size. I've got 4 uca-36 so I'll have to make a quad of mic preamps and then record some drums to compare to other solidstate pres.
Matt.
 

Attachments

  • uca-36.JPG
    uca-36.JPG
    95.5 KB · Views: 28

Latest posts

Back
Top