[BUILD] GIX-51X tube preamp

[bruce0]

Success. I am happy with the now undetectable (to me) noise levels in the HV power supply now, and I am in the process of putting all the modules together the same way.

I am using the 330uH power inductor (L501) with the .250 sense resistor as originally spec'd.  ( See note below on Inductor tests I ran if interested), replacing L503 with a bourns 1200uH inductor and a .1uF cap on the back of the board to ground. I am using 200 Ohms as a HV Jumper, and I am putting shields on the HV board, because I think I heard the difference, but I am no longer sure and I am venting them because the get mildly warm and there is an electrolytic in there. I have updated the previous post with my final changes.

The preamp is quiet, and as I have been doing testing and the differences I hear are from individual noisy tubes, and I have replaced a couple and all 4 modules are quiet now.

I have attached an updated BOM to this post with the parts I used.

bb

Note:  Leaning inductor L503

Note: Tack solder .1uF on back

here are more details, if you are interested, in my testing results.

Inductor Tests
I have now tested 470uH and 390uH inductors with the HV Board as L501 (the power inductor).  They both work using 0.200 ohm sense resistors.  However they run at 99khz, and 70kHz respectively and I think that is too low, and I think the 330uH is the one I am going with.  I think I want the switching to be further from the audible frequency range, and easier to filter.

I did some stress tests by putting a 47K resistor to ground at the driver stage plate, as a way to test these HV power supplies with higher current.  (The 470uH works with 0.250 resistor, but add 5ma through that 47k resistor, and it starts to sink to lower voltages. They do work with the higher current with the 0.200 ohm resistors).

I like the 330uH inductors best, with the 0.250 ohm sense resistor.  I have test it audibly, and with a dummy mic, and gain and output all the way up, it seems quieter, but I can't really tell.

I did some tests on the filtration for noise getting back on the 24V input power line, and the noise that gets through that filter is below my ability to measure.  So I am not worried about the switcher dirtying up the power for the other modules. 

Filter Inductors
As to filter inductor L503, I have tried filter inductors with a Q factor of 20 and of 90, and they all work the same as far as I can tell, they all will fit but they need to lean away from the HV board especially if you fit a shield on it. I have tried the Murata and the Bourne 1000uH inductors.  I have the others, but I am happy with the Murata (I am getting the noise on the HV after filtering at -80dB (about .007V rms noise on 252V, .005V pk-pk).  I can measure the noise in the drive stage (but it's not audible) so I am upping my resistor to 200 Ohms in the HV jumper.

I think any of the following will work for the L503:

Manufacturer            Mouser

(I tested) Bourns      652-RLB9012-102KL      RLB9012-102KL    Q>20
(I tested) Murata      580-18R105C            18R105C          Q>92
(I tested) Bourns      652-RLB9012-122KL      RLB9012-122KL    Q>15 - a bit bigger

I inspected an example of the following also, they fit, I think they would work fine.
          Abracon      815-AIUR-02H-102K      AIUR-02H-102K
          Fastron      434-01-102J            434-01-102J 

The cap to ground on the back of the board with this inductor is:
Murata      81-RDER72J104K8M1C1A    RDER72J104K8M1C11A  Multilayer Ceramic Capacitors MLCC - Leaded .1UF 630Volts 10%

 

[irfrench]

bruce0,

That's great news, thanks again for your perseverance.  I'm placing a mouser order for a couple other projects today so this is great timing.

If you don't mind a daft question and brief summary:-

L501  - same as BOM
.250 sense resistor - same as BOM
L503 - now 1mH
additional 0.1uF cap to rear of board as per your photos

When you say HV jumper what do you mean?  Sorry for the dumbness.


Thanks,
Ian

 

[bruce0]

HV jumper is a wire jumper that Rodney included in case the G9 needed more HV psu filtering.  I thought it did.  It can be seen below.

 

[irfrench]

Ah, great thanks for that.  Will any ol' 200ohm resistor do or do we need >.5w?

In a glorious turn of events I've just received my parcel of GIX boards!

Thanks Rodney, these are GREAT!


Thanks,
Ian

 

[bruce0]

i used a 1/2 W resistor.  But the voltage drop across the resistor in the idle state is less than 3 of volts 249V on one side, 252V on the other.  Jacob specifies a 2W resistor for R31, which has a 5V drop across it (250V/245V).

I would have thought that 2W was way overkill ( 3V over 200 ohms is 15mA *3 is .045W! And R31 5V over 1000 ohms is 5mA *5 or .025W)  but maybe there is a vacuum tube notion I am not familiar with.  Or maybe i should crank out a huge signal into 600 ohms and remeasure. 

So the answer is, you got me!

Maybe some Valve guy would chime in.

 

[gemini86]

There's really no other trick here. The resistor just needs to hold up under max current draw. I have yet to measure what that is, but once i get around to adding the filtering to my unit, I'll report back.

 

[bruce0]

Correct.  But the cap on the back really matters, if forms an LC filter with the 1000uH cap that causes a steep cutoff (with no peak because of the resistor) at about 1k and makes the biggest difference.

I added large caps and such as listed in the other post, but that was just to try to replicate the caps on the g9.


If you want i will send you (gemini86) a little shield and you can see what you think.  I will PM.

 

[tonycamp]

Bruce, thanx again for your work on this, i will be ordering up some mousakey, and getting started as soon as i get a minute.

 

[vineyardgray]

Am I the only one who gets very bizzare formatting (i.e. no formatting) with Bruce's Rev. 6 bom?    If not too much trouble, could someone who's local copy isn't all mangled by Notepad post a pdf?

Thanks Bruce for your hard work and Rodney for the great project! 

 

[Winetree]

+1
It's Better than nothing but will take a lot of time to unscramble.

 

[bruce0]

I will post a pdf, sorry.  GROUPDIY did not allow excel uploads (maybe it is not a open format?)

Attached is as PDF for your viewing pleasure.

Cheers

BB

(PS:  If you "FILE->IMPORT" that the original into almost any spreadsheet and choose tab delimited file and you will have a readable and editable spreadsheet).
Note the BOM is not complete, stuff is missing mostly if i had it

 

[irfrench]

Hi Bruce,

Just for clarity:

"R504 + R503    A single 1800 ohm resistor (measure these with a DVM, a small difference on
this part makes a big difference in voltage)"

Does this mean that you used a single 1800ohm resistor to bridge where R504 & R503 would have been?  I think the word 'these' has thrown me.

I've attached a .png showing the location of the 1k8 in RED.  R504 and R503 should be left out. 

If you could also repost the picture of the cap on the underside of the board that'd be really helpful.

Thanks again,
Ian

 

[gemini86]

That is correct. the cap goes between the output of l503 and ground.

 

[bruce0]

(Corrected resistor numbers) R503-R504-R502 form a voltage divider between the output of the boost converter and ground.  The feedback pin (FB ) on the Max1771 wants to see 1.5V for regulation coming from between R502 and R504.  It is set up so you can combine two resistors to get there (R504 and R503). You can use whatever resistors you want to get to that value.

I just happened to have 300K and so I needed 1800R to get to 252V  and I had one, so i jumped over two resistors with one.  But if you have a 270K resistor you would use a different value which you might make up from two resistors.

The factor of output voltage seen at the R502 R504 Junction is (R504+R503)/(R502+R503+R504)
In my case:
1800÷301800 = 0.005964215

The regulation target voltage is when the FB pin is at 1.5V so:
My regulation target is:
1.5 / 0.005964215 =251.49 (Volts)

But if you happened to have a 270K resistor and a 1K resistor you can "cross multiply" (sheesh) 1800 X 270000 /300000 =1620 so you would need about 620 ohms.

1620÷271620 =  0.00596421471
(1.5 volts) / 0.00596421471 = 251.5 volts

The most sensitive parts are the small resistors a small difference in value makes a big difference in output voltage, so it is nice that the card has a spot for two resistors. 
Overall you want the total resistance of the three resistors to be large (they are across 250 volts or so)

 

[irfrench]

Now I see!  Thanks.

So the originally spec'd R503 and R504 are used when in combination with the originally spec'd R502 (270K).  I'll measure and calculate before install and aim for 250v.

Thanks again,
Ian

 

[bruce0]

Exactly, and you got the correct resistor numbers (I corrected my post).  When the unit is running it drops about 2 or 3 volts across the HV jumper if you use a 200 ohm resistor there.  I used 350V caps because the voltage goes a bit high till the tubes get hot.

( if you are using 250V caps shooting for 250 will exceed the rating for 20 seconds every time you power up)

 

[bruce0]

Note:
I am getting Pops when changing Gain.  The G9 BOM recommends MBB (make before break) switches, also known as shorting switches.  The switch in the BOM is non-shorting.  I plan on trying to replace mine, will report the results.  I am hoping that the backplane of the switch is the same so I can just unsnap the front rather than unsoldering.

The pop's are not bad, I thought they were tube microphonics when they sat on the bench because the unit moved with each click, but in the rack they are audible.  Because of the clever gain structure that shifts from first stage to second stage I am not sure a cap across the anywhere works to minimize this, I will try to look into it.

In any case I recommend choosing but NOT 105-2454 which is non-shorting.

 

[irfrench]

Thanks,

I'm having a crack at getting this near completion tonight.

When selecting R503 and R504 are there any value limits I should avoid.  I've populated the board with 120R and 1k5 1/4w resistors as this seemed the simplest solution to getting the 1620ohm (with 270K - R502) - when supping my coffee at work it dawned on me that I hadn't thought about the possibility of a too low value at 1/4w.

Hope that makes sense?

Thanks,
Ian

Current state of things:

 

[bruce0]

Ian:

At 250V a 250K resistor will run 1 mA, so 1ma times 250V is 1/4 watt.  Don't go below 250K total resistance.  But the real issue is not too low a value because of resistor rating, the feedback pin (FB) needs like 20 nano-amps (I.E irrelevantly small), the problem is that the resistor divider is also a LOAD on the output (since it is from the output to ground). 

The 2 valves use about 12.5 of B+ (I can't calculate this, but this is just by measuring the voltage drop across the filter resistors I put in).  This means that if the voltage divider is in the order of 50K or less you are adding 50%! to the load on the power.  SO you want a large value resistance here.  I wouldn't go below 250K total resistance just because the load might effect the heat etc, and I have tried values up to 2.2M and they all power the unit (though at the time I was not measuring results properly, see earlier posts).  Earlier we thought a higher resistance might have interfered with the regulation but I am of the opinion that anywhere between 250K and 12.5M is ok (at 12.5M there are 20,000 nano amps running through the network).  So look around the resistors you have and find a combination that gives you the voltage you want.

bb

 

[irfrench]

Bruce,

Thanks.  Currently I have installed:

R502 - 270k
R503 - 1k5
R504 - 120R

All 1/4w...

Thanks again.
Ian