Late 50' tube... who guess ! (The Ondioline)

[zamproject]

The schematic doesn't show the output transformer secondary is grounded

yes, I check this 3 weeks ago, have some read, and still don't know if I have to ground it.

I'll measure tomorrow what's the peak to peak at speaker, and see what's requested to not fry my AD input...

Regarding hum ,"virtual" ground at heaters can be an option, but I have to check closely the whole ground wire.
Basically the ref wire (solid coper) run soldered at each frame all around, I see it like a ground plane using the 3 chassis...
I add an earth at main input (for safety), at the PSU/AMP section I post schemo today.
Then an umbilical go to the oscillator section (7 tubes, OSC/filter/tremolo/subOsc) with only 3 wire, AC heater (2wire) and B+
Then a second umbilical (from OSC section/frame) go to the keyboard and oscillator fine tuning capacitor at the upper section of the cabinet. There is a global pot volume (knee activated) which weeper go back (shielded cable) to the first block (PSU/AMP) and that's the only ground return to the first block for the whole "machine"
So the heater is floating in the upper oscillator frame (only referenced via umbilical to the grounded trafo 6.3 secondary at first block)
And all ground come back to the AMP via input banana jack

Best
Zam

 

[abbey road d enfer]

Top to bottom the pot measure 700k (for a 500k label...)
I read 79K at the wiper so just over 10%
I should probably better use a 100k pot in series with a 500k (or 200k with 1M) to optimise fine tuning accuracy ?

Yes, something like that. Now you got the gist of it, make your own sauce.

Anechoic is more a joke, but by listening in my "quiet" room I'm sure part of the background noise was also due to the transformer noise mechanical coupling (which was close), as I say I use a B&K (omni) which pick anything
I definitely have to improve hum, when I check/set THD at low frequency (close to 50Hz or upper harmonics) intermodulation beat a lot. 50Hz is 30 db below signal when I set THD... louder than h2 and h3 and with a noise about -60dB it's tricky to adjust harmonic peak in all that noise.

I didn't realize you were actually making acoustic measurements. As suggested by trobbins, you should make electrical measurements.

 

[trobbins]

It sounds like grounding will be a compromise. I'd recommend making a fixed protective earth link from the PSU chassis to other metalwork. Then it depends on any cable screens that also connect to chassis. For starters it may be worth your while making a grounding sketch showing what is grounded and where (as in a block diagram). Some care is needed to sort out possible hum loops - it took me a while to work through a Selmer organ from 1957 that I restored.

 

[zamproject]

As suggested by trobbins, you should make electrical measurements.

Any recommendation regarding resistor value in parallel of the speaker ?
I have 10k and 1k both 11W laying around.

 

[trobbins]

Some aspects are the output impedance of the OPT (eg. speaker could be 8 ohm) as any testing would typically use a resistor load of that impedance. Another aspect is the input resistance of your soundcard setup - some soundcards just have low resistance inputs like 10k, whereas others can include an un-balanced input (typically 0.5 to 1.5 Meg). Also the approximate output power of the amp stage needs to be estimated so that the max likely voltage across a resistor load can be calculated (eg. if 10W was expected in to an 8 ohm resistor then about 10Vrms is anticipated max output signal).

Adding a resistor divider can simply be done with 0.4W leaded resistors if the power in to those resistors is low enough, and if the soundcard input resistance does not significantly alter the divider ratio. A divider with 1K:100R could consume up to 100mW, so 0.4W resistors are ok and the divider won't load the 8R load on the output, and an example 10k input resistance from soundcard would not significantly alter the divider ratio, and the divider output could be up to 1Vrms, which should be ok for many soundcards to manage.

The absolute divider ratio is not really a concern, as any measurement would need to measure rms voltage across the speaker load and relate that to the reported soundcard/software level, as well as measure the speaker loading resistance (the accuracy of which is mainly determined by say the 8R test resistor which has to handle at least say 10W and not significantly add its own inherent distortion).

 

[zamproject]

Thanks for clarifications Trobbins

In the mean time I looked at my marshall 2550 schematic because it have a DI which is picked at trafo out, the divider is 27k/680
Beside the voltage divider they add a filter for frequency response correction.

I was able to set first harmonic 65dB below f , in the 80-200Hz range
This is 15 dB better than previous method (acoustic)
I'll standby this THD/poweramp topic and concentrate to other optimisations requested, I'm on the hum/parasitic noise today

Best
Zam

 

[zamproject]

Update of the month...

I moded the heater 0V/ground ref from one side trafo at chassis to floating with two 100ohm res, I think it improve a little the hum
OT is now grounded
The "corde" percussion system is now operative without any link to B+... that was one of the main safety issue with the syth, this copper string was directly connected to B+ via a simple resistance voltage divider.
I mod the touch detection system I design for my fader automation few years ago, now drive a reed relay that switch the 6BA6 g2, work great and safe !
I had big issue tuning the syth, especially the temperament tuning (octaves width), due to various leak and capacitive coupling in the oscillator circuit, this was very tricky but now I'm not that far from good result (considering a 70years old design)
Still I think I can optimise a step further, by changing some support material.

I'll probably do a special arrangement/cover song as soon as this restoration is finished (soon)

Cheers
Zam

 

[zamproject]

Well...bad news of the day...Ondioline release smoke today

Not sure yet what was the issue, I turn off as soon as I heard some crackles.

The -joke- being I was in the process of finally measure and take precise notes for ALL voltage value and current consumption (to have full data for trafo spec...) The mistake probably was to insert my DMM in place of the fuse to calibrate it for best value (I previously place a 400mA) but was on 10A setting at DMM so not that much chance to blow fuse before smoke

I test the PSU/power amp block with all tubes removed, all secondary AC ratio seems ok with 24VAC at main (variac)
But if I push too much it's like someone is eating chips inside, so there is probably some short.

That hurt, especially after 6 month half time on it, while the syth was almost finished and sound F... awesome
I'll start to look at possible replacement...

Zam

 

[trobbins]

That's a tough day on the bench. At least you know the Ondioline intimately now, so step by step testing can hopefully isolate the problem fairly rapidly. I'm pleased to hear that you were aiming to optimise the fuse value/type - few people do that, or even try to assess what fusing is appropriate, which seems such a basic step in alleviating collateral damage to their equipment when a part eventually fails.

 

[zamproject]

At least you know the Ondioline intimately now,

that's for sure !

For now I think the problem is some winding isolation
at low voltage (10%) all secondary read correct
for 24VAC pri, I have 660mV at 6.3, 50V for HT and 520mV at 5V (all unloaded)
I think I don't have any short, and I read around 5Mohm isolation between pri and sec
So my guess is some insulation material get deteriorated by time (like varnish around copper winding and cardboard separator...) especially since I power on the Ondioline probably tens of hours past two month, things get hot and the 70years old trafo fail...crackle start at about 100VAC pri, like arcing between non isolated and too close conductors.

I already pick two possible replacement at Hammond and Edcor, still I have to do some math to be 100% sure
I'll also try to open the trafo and see if something is possible to repair (including rewinding...) if the assembly allow it.

My Ondioline use
1 5Y3 rect
1 OA2 reg
2 6F6 (power)
1 6BA6
1 6J5
6 12AU7

Best
Zam

 

[trobbins]

Can you confirm you isolated all the PT windings except for the mains connection, and then used the variac to bring up the primary winding?

Imho the next test to indicate if the PT was a concern would be (with all the wiring to the PT windings isolated) an insulation resistance test to check for leakage at up to 1kVdc between windings and also to core. If you still have circuits/wires/parts connected to PT windings then they may also be suspect.

 

[zamproject]

Hey trobbins

No, I can't say I totally isolate the PT, I only remove tubes, which let some floating connections (except from one side of the 5V)
Unfortunately now it's too late, I completely dismantle the trafo...all laminations are off and already external 5 and 6.3 winding.
I had run a last test before that and clearly see the smoke from one core side, probably the HT winding (which is next to remove but I need to build a small tool to reel up the thin wire)
I have no tool to generate 1kV...

I'll probably take a Hammond 370FX, which should be ok, maybe a little to low on the HT but as input is multitap I can set it from 200 to 240 within 10V step

Best
Zam

 

[abbey road d enfer]

I'll probably take a Hammond 370FX, which should be ok,

370FX is rated at 550V center tapped, i.e. 275V AC, which translates in about 350-360V DC after rectification. The amp expects 280V...
I believe a 369HX would be more adequate.

maybe a little to low on the HT but as input is multitap I can set it from 200 to 240 within 10V step

It's very dangerous to overvolt a xfmr. What's more, it often results in dramatic increase of hum.

 

[zamproject]

Hello Abbey

Aren't the trafo spec given unloaded ?
Before smoking, original trafo give 580V unloaded at HT (290-0-290) (with variac at 230 and trafo at 240 setting)
Yesterday at 10% pri (24V) I get 60V sec, so 600V, witch is consistent regarding ratio
It effectively give 280V after 5Y3-C-L-C with all tubes
Choke measure about 200ohm but no idea of inductance.
I previously take lot of measure, but I don't have the Vsec loaded and V at rectifier in my notes
so not sure of the effective voltage drop...

I believe a 369HX would be more adequate.

150mA should be ok too it's only a duet PP but if I understand correctly the OA2 can pump about 70mA at startup
so I should probably be careful to not use too low current capability trafo.
Heater is about 4A for 6.3, so 5 or 6 is fine, 5V@ 3A is fine too (2A for 5Y3)

Also I try to find something in stock... don't want to wait 5 or 10 weeks

It's very dangerous to overvolt a xfmr

in practice I was more thinking "real" 230 setting as that's what I have here.

Best
Zam

 

[zamproject]

Aren't the trafo spec given unloaded ?

my mistake....hammond specified -under load-...which printed vertical in the datasheet make me read -unloaded-
tks dislexia

 

[abbey road d enfer]

Hello Abbey

Aren't the trafo spec given unloaded ?

As you've found out, no.

Before smoking, original trafo give 580V unloaded at HT (290-0-290) (with variac at 230 and trafo at 240 setting)
Yesterday at 10% pri (24V) I get 60V sec, so 600V, witch is consistent regarding ratio
It effectively give 280V after 5Y3-C-L-C with all tubes

It seems to indicate something is not up to par, probably the valve.
Anyway, the 6F6's would stand more than 280V, but the capacitors may not. Have you checked them? I would think they're the most likely to fail, and may have taken the xfmr with them.

so not sure of the effective voltage drop...

With 2x275Vac the graph for 5Z4 gives a DC output of 325V, to which you substract 10 V in the choke, resulting in 315V for B+. This is for 80mA quiescent.
If it's actually a 5Y3, it has higher voltage drop than 5Z4, so it may be correct.

150mA should be ok too it's only a duet PP but if I understand correctly the OA2 can pump about 70mA at startup

I don't see an OA2 in the schemo...

so I should probably be careful to not use too low current capability trafo.

Always a good option.

 

[zamproject]

ho sorry Abbey

I think I don't post the second part of the shemo...

Anyway, the 6F6's would stand more than 280V, but the capacitors may not. Have you checked them? I would think they're the most likely to fail, and may have taken the xfmr with them.

6F6 are on the high side with about 14.7V in my notes, across the 250ohm cathode resistor which is close to 30mA per tubes
All alu have been changed with 450V or 500V caps, they have less than 50h operation

I just finish the HT unwinding at trafo (about 800 turns ) , last end is burned at junction to the external wire, and I suspect I smoke pyralene that is probably used at kraft paper winding separator, they seem well "oiled"

Best
Zam

 

[zamproject]

OK, issue 100% spotted
the assembly concentric winding are as follow, from inside to outside, main-screen-HT-5V-6.3
The 3 turn of thin isolation paper have burned between screen and first HT end.
The wire that connect the thin winding to external world, twisted and soldered together, probably create an extra thickness that mark the isolation by the time until they finally touch or get really close resulting in arcing...

That was more or less what I expected, but know I have a clear visual explanation.

I think I'll be able to fix this but not to rewind the PT, and it's probably better to buy a new one anyway, way safer, there is too many paper, cardboard and even wood inside this trafo (wedge to compress the E shape lamination and the winding assembly)

Best
Zam

 

[zamproject]

just to add

I remove the 4 turn of kraft paper after the thin copper sheet for screen (which was perforated by arcing...) and only one layer was not perforated too before the main winding

also in my syth I have to more 12au7 (sub OSC) that are not in the initial schemo posted above.

Cheers
Zam

 

[abbey road d enfer]

ho sorry Abbey

I think I don't post the second part of the shemo...

The OA2 is fed via a 100k resistor, so the current could never exceed a couple of mA.