DaveP

Analysis of Altec 436/438
« on: February 19, 2010, 09:56:31 AM »
Hi Guys,
Sad git that I am, I worked out how the threshold control works and some implications which may interest some of you with curious minds, here we go.

If we start with Altec's published figures of 0dBm, +10dBm and +16dBm on the compression chart, we get the following voltages across a 600 ohm load, 0.775, 2.45 and 4.88 Volts.

Now thanks to CJ's "take apart" http://www.groupdiy.com/index.php?topic=35658.0 we know that the turns ratio on the output TX is 6.2:1, this means that our voltages are now 4.805, 15.19 and 30.26 on the primary.

These RMS voltages appear across both 270k resistors so we must divide by 2 and multiply by 1.414 to get the following peak voltages on the cathodes of the 6AL5; 3.40, 10.74 and 21.39 Volts.  These are the voltages available by turning the threshold control through its range.  By adjusting the threshold control, the 6AL5 only conducts when the peaks exceed the set voltage, it's a very elegant system.

Now for the interesting stuff on the other models; the 436A has a voltage divider on its HT to provide 10.6 volts, so this model is set at +10dBm with no adjustment (10.6 is within the spec of 10.74V above).
The 438A does not have the divider but connects the threshold to the cathodes of the 6CG7 at 2.3 Volts, so this model is fixed at 0dBm threshold.  Please note that the HT supply to the 6CG7 is 145 volts on these models so the cathodes are at ~2.3 volts.  The 436C has an HT of 150V which in fact pushes up the cathode voltage to 3.4 volts (I measured it) so again the lowest setting is dead on 0dBm threshold.

Two last points, The Sowter 9745 Transformer for the 436 is only 5:1 so this changes the threshold/gain slightly, but as long as you are aware of this and allow for it, it is no deal breaker and it works a treat.

PRR mentioned in an earlier post that he thought silicon diodes were an ok sub for the  6AL5 but I found this post, http://blogforumpost.blogspot.com/2009/04/theory-of-vacuum-tubes-5.html which states that the 6AL5 has no diode drop.  This means that silicon diodes would have a voltage drop of say 1.4V which would alter the threshold voltages around the 0dBm mark by quite a lot and by a tad right through the range.

I hope you guys find this interesting, now what did my wife mean by "get a life" hmmm.
all the best
Dave P

Just turned 60 and getting younger by the day!
Soundcloud: Delayed Action.


opacheco

Re: Analysis of Altec 436/438
« Reply #1 on: February 19, 2010, 07:27:06 PM »
wow!!...Dave P, that's very interesting caculations!!...., Could you be kind in order to explain the "By adjusting the threshold control, the 6AL5 only conducts when the peaks exceed the set voltage" process please???. That's is no so clear for me ...Sorry.

Thanks a lot,
Opacheco.
I love the Vaccum Tubes Sound!!!

PRR

Re: Analysis of Altec 436/438
« Reply #2 on: February 19, 2010, 11:08:34 PM »
> 145 volts on these models so the cathodes are at ~2.3 volts.  ... 150V which in fact pushes up the cathode voltage to 3.4 volts

Must be something else also. (I don't have alll variants memorized.) 145-150 does not explain 2.3V-3.4V. Different cathode resistor? Bleeder current?

> 6AL5 has no diode drop.

Over-simplified.

Pull 100mA(!!!), big forward drop. It may pass 1uA at "negative voltage". Have to put it in context.

> silicon diodes would have a voltage drop of say 1.4V

Say 0.6V, more or less. Just one diode-drop in the path, and current is small.

> peak voltages on the cathodes of the 6AL5; 3.40, 10.74 and 21.39 Volts

Thanks for working the digits.

Taking a dart-guess that going from 6AL5 to 1N4007 makes a 0.6V difference, and that 1dB difference is small, if threshold is 6V or more, hollow/solid diode makes "no difference". This works for the 10.7V and 21V condition, but does fail at the 3.4V condition.

For thought: What happens above threshold? I bet you find that 3V thresh veers into a soft rising "limit", while 21V breaks sharp and flat. If we want sharp/flat (we may not), then we should have a sidechain amp, not rely on line-amp level. OTOH, the economy of tapping the line-amp is unbeatable, and these were hardly "high end" boxes like the big GE and RCA film and radio limiters.

DaveP

Re: Analysis of Altec 436/438
« Reply #3 on: February 20, 2010, 06:54:52 AM »
(Could you be kind in order to explain the "By adjusting the threshold control, the 6AL5 only conducts when the peaks exceed the set voltage" process please???.)

OK Opacheco, looking at the schematic http://www.freeinfosociety.com/electronics/schemview.php?id=1331
The threshold control is P3 the 25k pot.  It makes a variable potential/voltage divider across the HT at the junction of R3, R4 & R13 which is at 210V with no signal.  The 270k resistor R15 drops the voltage down to about 20V as marked on the schematic.  So by varying the threshold control the wiper can sweep from 2.3V up to 20V.  The wiper is connected to the 220k resistors and they are connected to the cathodes of the 6AL5.  The two 220k resistors are necessary to provide a big enough AC load to the 6CG7 so that it won't drag down the output voltage too much, the capacitors C5 and C6 stop the DC on both sides from meeting but allow the AC to get through to the 6AL5.  The setting of the threshold voltage sets the working bias of the 6AL5 so any peaks over the set point get through the diode and get rectified to a negative control voltage for the 6BC8.  These capacitors are carefully chosen to provide the right level of low bass cut so that the amplifier does'nt "motorboat" with feedback through the power supply, but thats another issue.

PRR, yes it is over simplified and I know the source resistance has a bearing here.  Altec say the attack time is 50mS but the attack resistor is only 33k not 50k so there must be 17k of source resistance in there somewhere if the spec is accurate?

Yes 1.4v is probably too high and we could put some Schottkys in there and get down to 0.5v but I guess the main reason they stuck with a tube is so it could be easily taken out to provide a straight amp conveniently?

Regarding the cathode voltage, I take your point and it could be a typo or meter error on their part and it just kind of stuck, but it is definitely 3.4V not 2.3V.

My main interest in this compressor is that the Beatles used them (in RS124 form) along with a Fairchild for most of their recordings, so they must have some virtue.  I like the simplicity of the audio path and the economy of parts as you say.

Incidently, I put an ECC82 into the 6BC8 slot (changed heater connections first!) as you had used one in your project.  Had a very interesting result, it had similar gain reduction to the 6BC8 but instead of the chart levelling off it continued to go south and curved back on itself, the difference between sharp cut-off and remote I guess.
all the best
Dave P
Soundcloud: Delayed Action.

gary o

Re: Analysis of Altec 436/438
« Reply #4 on: February 20, 2010, 07:45:49 AM »
Wow this is great reading for me Im very interested in Vari mu and the 436 in particular I find it hard to understand electronics sometimes especially the maths, so as I say great reading.....Im waiting for a 6BC8 to arrive so I can knock a 436 together to experiment with I have all other parts waiting, as Dave says EMI believed in this little circuit.....I just did a little point to pint knock up of the VK1 circuit thats similar to the 436, I have been wondering about the side chain in this circuit .....would anyone care to take a look and explain this design please, I read that its wrong but seems to work well to me ..

Here is link Im hoping will take you to schematic

http://purusha.smokinggunrecording.com/XVXDY/1schems.pdf

opacheco

Re: Analysis of Altec 436/438
« Reply #5 on: February 20, 2010, 09:41:13 AM »
(Could you be kind in order to explain the "By adjusting the threshold control, the 6AL5 only conducts when the peaks exceed the set voltage" process please???.)

OK Opacheco, looking at the schematic http://www.freeinfosociety.com/electronics/schemview.php?id=1331
The threshold control is P3 the 25k pot.  It makes a variable potential/voltage divider across the HT at the junction of R3, R4 & R13 which is at 210V with no signal.  The 270k resistor R15 drops the voltage down to about 20V as marked on the schematic.  So by varying the threshold control the wiper can sweep from 2.3V up to 20V.  The wiper is connected to the 220k resistors and they are connected to the cathodes of the 6AL5.  The two 220k resistors are necessary to provide a big enough AC load to the 6CG7 so that it won't drag down the output voltage too much, the capacitors C5 and C6 stop the DC on both sides from meeting but allow the AC to get through to the 6AL5.  The setting of the threshold voltage sets the working bias of the 6AL5 so any peaks over the set point get through the diode and get rectified to a negative control voltage for the 6BC8.  These capacitors are carefully chosen to provide the right level of low bass cut so that the amplifier does'nt "motorboat" with feedback through the power supply, but thats another issue.

PRR, yes it is over simplified and I know the source resistance has a bearing here.  Altec say the attack time is 50mS but the attack resistor is only 33k not 50k so there must be 17k of source resistance in there somewhere if the spec is accurate?

Yes 1.4v is probably too high and we could put some Schottkys in there and get down to 0.5v but I guess the main reason they stuck with a tube is so it could be easily taken out to provide a straight amp conveniently?

Regarding the cathode voltage, I take your point and it could be a typo or meter error on their part and it just kind of stuck, but it is definitely 3.4V not 2.3V.

My main interest in this compressor is that the Beatles used them (in RS124 form) along with a Fairchild for most of their recordings, so they must have some virtue.  I like the simplicity of the audio path and the economy of parts as you say.

Incidently, I put an ECC82 into the 6BC8 slot (changed heater connections first!) as you had used one in your project.  Had a very interesting result, it had similar gain reduction to the 6BC8 but instead of the chart levelling off it continued to go south and curved back on itself, the difference between sharp cut-off and remote I guess.
all the best
Dave P

DaveP,...Thanks a lot for your explanations!!!...in conclusion: the R15 and P3 is a Bias polarization network for the 6AL5 taking the 210volts HT supply for the in the 6BC8.....right?.
In DC analysis this become a FEEDBACK network (that's look to me!) coming from the 6AL5( its equivalent network will be a kind of DC source changing in proportion to the output Ac voltage) this in turn will change the Q point of the 6BC8….Am I wrong or that’s true???...for another hand, the 6CB8 and the 6CG7 are working in the diferential mode…right??
Thanks a lot,
Opacheco.
I love the Vaccum Tubes Sound!!!

DaveP

Re: Analysis of Altec 436/438
« Reply #6 on: February 20, 2010, 12:46:30 PM »
DaveP,...Thanks a lot for your explanations!!!...in conclusion: the R15 and P3 is a Bias polarization network for the 6AL5 taking the 210volts HT supply for the in the 6BC8.....right?.
In DC analysis this become a FEEDBACK network (that's look to me!) coming from the 6AL5( its equivalent network will be a kind of DC source changing in proportion to the output Ac voltage) this in turn will change the Q point of the 6BC8….Am I wrong or that’s true???...for another hand, the 6CB8 and the 6CG7 are working in the diferential mode…right??

I guess you are a solid state guy trying to understand tube circuits?
Its probably more simple than you think, the negative DC output from the 6AL5 adjusts the grid voltages of the 6BC8 changing its gain instantaneously with the output voltage.  It varies from 0V (maximum gain) to -10V (minimum gain).  You are right in that the 6BC8 and the 6CG7 are working in push- pull (differential mode) in order to cancel the considerable distortion that arises from operating the 6BC8 in this way.  The halves of the tubes have to be very well balanced to avoid "thumps" (low frequency noise) from this type of circuit.
all the best
Dave
Soundcloud: Delayed Action.

eskimo

Re: Analysis of Altec 436/438
« Reply #7 on: February 20, 2010, 12:54:33 PM »
Excellent stuff guys!

DaveP

Re: Analysis of Altec 436/438
« Reply #8 on: February 20, 2010, 07:22:12 PM »
would anyone care to take a look and explain this design please, I read that its wrong but seems to work well to me ..

Hi Gary O,
I looked at the VK1 schematic and it has moved a long way from a 436 has it not?  Not sure that a 12AT7 is up to the job, two doubled up would be, but I noticed that the output has been beefed up with DC feedback to the G.R. tube, big change.  There does not look to be sufficient resistance between the output tubes to me, only 2x 65k, this puts a big load on the output tube.  Anyways, the timing circuit is fairly straightforward and the performance can be calculated from the component values as PRR has done if you search the Vari-mu index.

Basically the attack time is the series resistor in ohms x the capacitor in F expressed in Seconds so you have C100 100nf x 220k max which is 0.000,0001F x 220,000 which is 22mS, this is adjustable down to nothing but you also have R112 68K and R12 15K in series with a switch making a second circuit with another timing which you can now work out as above.  The release time is calculated in the same way being (P2 + R9) x C100 making a release time variable from 15mS to 115mS, again this is complicated by C4 making a second slower release rate of 1.15 seconds down to 115 seconds.  The release is therefore designed to have a fast initial release followed by a slower release.  The attack control also interacts with the release control the way it is wired because the leakage path to earth includes P101, if the release control was on the C100 side of P101 it would act alone.  The thump switch is either there to add thump or remove it!  But if the push-pull tubes are properly balanced there should be no thump anyway to my mind.

Overall this circuit seems over complicated and a complete departure from a 436 and is very unlikely to sound like one because it doesn't have the same tubes or schematic.  But hey if it works and you like it why not?  Interesting to try it back to back when you make a 436.  I hope this has been of some help.
best
Dave P
Soundcloud: Delayed Action.

PRR

Re: Analysis of Altec 436/438
« Reply #9 on: February 20, 2010, 08:50:07 PM »
> states that the 6AL5 has no diode drop.

Some white-shirt junior engineer sweated over a hot 6AL5 to make these curves; don't let that info feel neglected.



The current may be "small" (consider a dozen-volt over-threshold and ~~33K attack resistance) but the voltage drop is non-zero. It fact it appears to lay on zero current for several tenths-Volt.

> Altec say the attack time is 50mS but the attack resistor is only 33k not 50k so there must be 17k of source resistance in there somewhere if the spec is accurate?

Audio-path attack time is not = to R-C time-constant. It is usually specified as time to get within a few-dB of the final gain, rather than the 63% change that R-C is specified at. Also: the grid law is curved, not linear. (Interestingly, these curves tend to result in an audio time similar to R-C time, and I usually assume they ARE equal-enough to guide a breadboard. Final timing must, of course, be by-ear or by-test.)

Added impedances: the source is a couple of triode plates. Maybe 6K plate resistance on datasheet, perhaps a bit higher where we run them, say 10K. Two on a PP winding act in parallel, 5K. This is in parallel with load, typically 1X-3X the tube impedance. Say 3K seen by sidechain. Some models have resistor dividers; that VK1's pots may add 8K. And as we see, each side of 6AL5 is about 500 ohms.


Val_r

Re: Analysis of Altec 436/438
« Reply #10 on: February 21, 2010, 05:02:59 AM »
> Altec say the attack time is 50mS but the attack resistor is only 33k not 50k so there must be 17k of source resistance in there somewhere if the spec is accurate?

It is usually specified as time to get within a few-dB of the final gain, rather than the 63% change that R-C is specified at.

Can you elaborate?
F.I., if we set a final gain of 6 dB of GR, how do we measure the attack time?
Would it be correct to state the time the waveform takes to reach -2dB (63% of -6dB, final gain red.) from 0 dB?

Ok, cool ! Jahmekya

gary o

Re: Analysis of Altec 436/438
« Reply #11 on: February 21, 2010, 05:32:10 AM »
Thanks for the VK1 investigation I cant take it all in as I have trouble with the maths Im afraid but very interesting..... I always thought the 436 and VK were very similar.....but not then ?.....It concerned me when I read that the VK1s side chain is a f*** up in the chaps own word & that it wouldnt work as the bias voltage is missing ?............I never seen the daul pot on any other vari mu ?

I have some notes that are said to be the RS124 mods so was wondering if the VK is a kind of modern version of modded 436......

Some interesting stuff its says the side chain coupling capicitors values were changed 3 times over the years...
The output valve grids were switched ? not sure what they mean ? if swapped would that flip the phase ?
Recovery circuit was taken off to lower distortion ?? what would that mean I wonder ?
Output feedback was added......Isnt that like the VK ?
Iput valve re biased ??
There was other modds listed more obvious practical ones if anyone wants I can list them, Im not sure if is really the EMI info but its interesting, I have other so called EMI modds but they def not as I found where they all came from.

Sorry for all the questions & sorry if anyone has answered my questions already as I say I have trouble understanding & the maths blow my mind its annoying as info is in front of me, Im really hungry to learn I really wanna build the 436 now to see how it sounds against the VK, the VK sounds way better than I thought it would I use it to record vocals I prefer it to LA2A .....I might build the 436 side chain & bolt onto the VK to see what happens..

Thanks for yr patience.

DaveP

Re: Analysis of Altec 436/438
« Reply #12 on: February 21, 2010, 06:27:56 AM »
Hi PRR,
thanks for your input, always good stuff...I had calculated 9K for the output tubes and you guessed 10k.  Some compressor specs come out exactly on the the 63% but I appreciate that another engineer might prefer 5 x longer and state 90%.

Gary O.
I am no mathematician, I don't call multiplication on a calculator heavy maths!  Only skill is knowing what your doing and where to put decimal point.  You have to master that to make a complicated bit of kit like a compressor, even a 436.  My advice is dump the VK1 if you have doubts and make a good 436 when you have the parts.

I am awaiting delivery of "Recording the Beatles" with supposedly, all the details of the RS124 mods, it will have cost me $160 to get that info and I think I will be sued if I posted the mods, but I would not want to anyway because they deserve the money for all the work they put into the research.  In any case I did not just want the RS124 details.
best
Dave P
Soundcloud: Delayed Action.

kambo

Re: Analysis of Altec 436/438
« Reply #13 on: February 21, 2010, 06:40:36 AM »

gary o

Re: Analysis of Altec 436/438
« Reply #14 on: February 21, 2010, 07:46:35 AM »
Thanks for postings I understand its not heavy maths I just trying to explain I have trouble understanding electronics specially when the numbers get crunched Its my problem Im just trying to explain why I cant always understand when things are explained but thanks for tryin .....because of this i build lots of stuff to try out for myself & its fun & I end up with good kit.... il keep my Vk1 as it sounds good and competes with my BA6A & as I say I have all the parts for a 436 just waiting for input tube... I have read many opinions on the 436 most of it negative sadly, except when Beatles are mentioned, personally I never really liked the beatles apart from hey jude I love that recording, I have 3 AKG C28A mics here that came from the BBC I wonder if there is beatle spit on the pop shields.

gary o

Re: Analysis of Altec 436/438
« Reply #15 on: February 21, 2010, 09:25:14 AM »
The 436 DC feedback part is that theR102 R103 100K resistors ? if so that would lower the output level further wouldnt it ? making it worse for the already struggling 12At7 ? why would the designer use this tube if not up to the task I wonder....

KrIVIUM2323

Re: Analysis of Altec 436/438
« Reply #16 on: February 21, 2010, 12:31:29 PM »
Hi everyone.

Quote
The 436 DC feedback part is that theR102 R103 100K resistors ?

Yes it is (on the VK1).

Quote
if so that would lower the output level further wouldnt it ?

Yes. Looking at 12at7 datasheet we can expect approximately 35db of gain under the operating condition in the final stage of Vk1 (ClassA 250V plate supply, 200 ohms cathode resistor it give an amplification factor of 60= so 35db of gain approximately). With feedback i suppose it give 5db off feedback  ( a figure of approx.17% feedback) which make sense for me.
Olafmatt supposed it was easier for stereo unit to be 'consistent' with specifications if i remenber well.

I don't think 12AT7 need to be doubled as 30db GR is big amount of reduction and the comp was probably not designed for such heavy treatements. But i don't know, maybe i'm wrong.

Quote
It varies from 0V (maximum gain) to -10V (minimum gain).  

I've got a question about this. I was asking myself for a long time now if the GR stage is working class A or B. Because for me with a grid at 0v the tube can't work in classA but in classB working with just one half of the input signal. Could someone explain me if i'm wrong and why. Thanks.

Edit: i forgot the cathode resistor R106 in my thought. Is it this one which bias the grid at a certain voltage and the sidechain wich sum with that bias to give the GR amount? If it is what resistor values are seen by each cathode? 110 ohm in the case of VK1?
« Last Edit: February 21, 2010, 12:55:39 PM by KrIVIUM2323 »

rotation

Re: Analysis of Altec 436/438
« Reply #17 on: February 21, 2010, 01:20:52 PM »
Hi everyone.

Quote
The 436 DC feedback part is that theR102 R103 100K resistors ?

Yes it is (on the VK1).

Quote
if so that would lower the output level further wouldnt it ?

Yes. Looking at 12at7 datasheet we can expect approximately 35db of gain under the operating condition in the final stage of Vk1 (ClassA 250V plate supply, 200 ohms cathode resistor it give an amplification factor of 60= so 35db of gain approximately). With feedback i suppose it give 5db off feedback  ( a figure of approx.17% feedback) which make sense for me.
Olafmatt supposed it was easier for stereo unit to be 'consistent' with specifications if i remenber well.

I don't think 12AT7 need to be doubled as 30db GR is big amount of reduction and the comp was probably not designed for such heavy treatements. But i don't know, maybe i'm wrong.

Quote
It varies from 0V (maximum gain) to -10V (minimum gain).  

I've got a question about this. I was asking myself for a long time now if the GR stage is working class A or B. Because for me with a grid at 0v the tube can't work in classA but in classB working with just one half of the input signal. Could someone explain me if i'm wrong and why. Thanks.

Edit: i forgot the cathode resistor R106 in my thought. Is it this one which bias the grid at a certain voltage and the sidechain wich sum with that bias to give the GR amount? If it is what resistor values are seen by each cathode? 110 ohm in the case of VK1?

35dB of gain only tells you "how much" tube amplifies. So this number can't simply tell you if tube is up to the task for, say, 20-25dB GR. It depends on output tube, SC and GR tube, it's all related and pretty complicated :-)
Btw, the problem with this output tube/SC is that it gives to small CV; from -8V to -10V, this is around 8-10dB of GR. Altec's GR tube works in similar was; you need around -1V of CV for 1dB of GR..
VK1 improves a little when i removed feedback resistors (r102,103), i get a little more compression.

You can read about single ended and push pull circuits here:
http://www.aikenamps.com/SingleEnded.htm
Class a/b and b are used in power amps where you want good efficiency. Preamps and compressors are normally functioning in class a, single ended or push pull (good cancellation of distortion and hum).

Miha


rotation

Re: Analysis of Altec 436/438
« Reply #18 on: February 21, 2010, 01:36:07 PM »
[quote author=KrIVIUM2323 link=topic=37878.msg466714#msg466714
Edit: i forgot the cathode resistor R106 in my thought. Is it this one which bias the grid at a certain voltage and the sidechain wich sum with that bias to give the GR amount? If it is what resistor values are seen by each cathode? 110 ohm in the case of VK1?
[/quote]

As i understand it R100,101 (grid bias?) are changing gm of the tube. So amplification is changing by varying gm.
Am i right?
« Last Edit: February 21, 2010, 01:39:27 PM by rotation »

DaveP

Re: Analysis of Altec 436/438
« Reply #19 on: February 21, 2010, 02:56:07 PM »
Hi Everyone,
Several vari-mu compressors use feedback, usual reasons are to lower output resistance and distortion and noise.  Could be VK1 compensated for lack of beef in 12AT7 by using feedback.  I have tried 12AU7 12AT7 and 6CG7 for this duty in a 436 and the 6CG7 is the best at 150V.  In fact it is quite instructive to look at the mu/gm/rp curves for these tubes at different voltages and then you can get inside the designers head to see why he chose that particular tube and that voltage, some decisions are down to cost but the characteristics of the 6CG7 are ideal in this circuit.  Its midway between a 12AU7 and a 12BH7 in terms of beef.  Having said that, if you want more gain then doubled 12AT7's work very well indeed.  A pair of those tubes will have half the rp and double the gm of a single tube.  This gives the doubled tube an rp of ~6K against 9k for the 6CG7.

As has been said before in the vari-mu postings (always good to check these out), its not really the mu varying that does the business.  Its the falling gm and rising rp that causes the loss of gain, the mu does fall but not as much as the other parameters change, just look at the curves on the data sheets, thats why PRR was able to use a 12AU7 tube in his compressor. Gain = muXRL/rp+RL so falling mu and rising rp are going reduce gain, simple!

The UA 175 and 176 use a 6BC8 GR tube and they are very well respected compressors.  If I wanted to make a better 436 then I would take some tips from the UA series, like regulated supply to the GR tube and the trimming pots etc.  A regulated power supply doesn't just fix the operating point, it reduces hum and stops motorboating through the power supply caps, a big plus!

Yes the GR cathode resistors are very important but are often screwed with when people start modding the meter circuit, but they should be left stock and a meter just put across the original cathode resistor together with a variable pot to zero if required.  It stands to reason that changing the cathode resistor will alter the operating point from the original circuit.
I'm glad my posting has sparked some debate.
all the best
Dave P
Soundcloud: Delayed Action.