SKYLINER build - inspired by ODS #124 and #060

[Stephen1966]

Some final thoughts about the build process... If you follow the layout after getting the boards in, it is next, a good idea to install the brown B+4 cable into the board with the wiper leg of the PI trimmer as soon as possible. If you're clumsy like me, the trimmer will get knocked around quite a bit while you're wrestling with all the boards' other wires and soldering in the B+4 cable provides the delicate pot's other legs with some stability. It's a short cable and it doesn't get in the way later on. This would be before it shows it installed in the current layout.

SAM_7581.JPG

I'll update and replace the layout and schematic and any other materials on the first page of this topic when I have confirmed my final component values.

[Stephen1966]

It is perhaps a bit different the way I will need to do it.

After forming, and at full voltage I''ll need to check on the dropping resistors in the power supply. I have my target node voltages but because the PS was only simulated and now we are in the real world I fully expect I will need to swap out some of the resistors for different values. I have my decade box to help with that. It's only really relevant with all the tubes in so it will be back and forth fast on fast off. I've done a preliminary test on the negative voltage at the bias filter and it's just a little over -60VDC which is on the money given there is no load yet and the tfmr tap is rated for 60V.

The start up procedures in circulation all seem to point to installing all the preamp tubes first and taking their voltage reading for confirmation, then it seems to be a good idea to turn the bias pot all the way down to its coolest before then installing the output tubes. Is this the time to remove the PI?

yes! ok for bias, you can put to -50V..
you have only one main concern to clarify now.. assuming that all is ok ofcourse, you need to assume something in the life from time to time you can go forward.

output tubes will oscilate or not ,after applying the 450V on them?? chances are 50-50% . oscillations are on full power , so load need to be connected for this stage!
there will be full powered loud noise on the speaker! if you dont connect the speaker and amp is oscillating you have bad chances to burn the output transformer (arc of few kV inside the layers in OT!) it is loud noise so the best to do this test when all your neighbours go to bed, @cca 23.00h

to reduce chances to 0%, you need to or remove NFB (without feedback oscilations will not happen 99%) OR to remove PI (without phase inverter, output tubes have no any loop closed).

without PI tube you can play with the amp all day, adjusting, adjusting the bias , measuring all voltages, without any trouble.

Good to know, thanks. I'm just thinking what else I would need to do if I swap the PI plates' wires, maybe swap the plate resistors as well? Hmm. I'm hoping it doesn't come to that. The output section just after the PI is all but equivalent either side of the PI but the grid and cathodes are pretty well baked in to the circuit with half the stage fed by its grid and the other half by its cathode . Is this 50/50 because of the OT windings or the PT? It was Andy Fuchs who mentioned something about reversing the HV leads from the PT to remove oscillation. Is this not a better solution to the problem than re-engineering the LTP?

If the neighbours complain, I just send the wife out. She's in PR so it's just like homework for her

[martin manning]

Perhaps someone could explain this to me?...

Here is a picture of the current limiting resistor installed prior to forming the caps....
I got a 140VDC voltage drop with this resistor, so today, I'll be forming the caps again without it, with full power.

The voltage drop didn't occur after the resistor however, but before it, at the tail of the rectifier.

all normal , you have bleeder resistor to the ground from the cap Ub+ to the GND, forming divider.. :P

If you tie a high value resistor between the output of the rectifier and 1st capacitor and measure at the rectifier, the DC voltage will not be smoothed at this point by a reservoir cap. So your DC voltmeter will measure the effective voltage of the non smoothed DC voltage.

Both of these answers are correct. The rectifier output will have a huge ripple riding on some DC, with p-p equal to the peak rectified voltage minus the smoothed DC. In your case the lower leg of the divider is about 150k, basically all the balance resistor series pairs connected in parallel. Measuring DC, about half(?) of the the voltage drop from the peak rectifier output to the smoothed DC voltage occurs due to the ripple.

The start up procedures in circulation all seem to point to installing all the preamp tubes first and taking their voltage reading for confirmation, then it seems to be a good idea to turn the bias pot all the way down to its coolest before then installing the output tubes. Is this the time to remove the PI?

I don't know why people like to start with the preamp tubes. I prefer to get the power amp running first, which brings the B+ voltage down near normal. Of course you need to connect a load whenever you power up with output tubes installed. I check the output transformer before installing it to confirm its turns ratio and phasing, so there is little or no chance of oscillation due to positive feedback. Some manufacturers show phase dots on the schematic, but there is always the possibility that the lead wire colors are reversed so I check anyway. Breaking the NFB loop or removing the PI tube allows getting the power tube bias in the right range first.

If you get it wrong and it squeals, swapping the phases at either the pi output or at the output transformer primary leads, whichever is easier, will fix it. PI plate loads stay as they are.

[Stephen1966]

Following the layout, the rest of the build goes fairly uneventfully. It's only at the end when installing the power switch and AC mains module that things get murky with the heater wiring. You may not use these kind of components so this is just what I did, but if you do you will want to wire the primaries of the PT and relay tfmr into the switch first, then one leg of the heater wiring soldered on the last tube socket, then your cables for the power switch from the mains module over the top of the black and white heater wiring. So that it feeds across the back of the switch as in this photo.

SAM_7734.JPG

If you try doing it in the order shown in the layout it's going to be difficult getting at the last tube's socket when it comes to the heater wires.

These heater wires, I'm not sure I like them but I'm sure if you run a heat gun over them twisted, all the loose wax flakes will melt into the cloth and make it firm prior to installation. Or you might follow Dumble and use something like AWG 18 PVC cable. His looks nicer if I am honest.

That's all I can think of regarding the build. Except don't forget to take lots of photos of every part of the process. It doesn't just allow you to publish stories like this one, but when it comes to checking the wiring of components later on when they are obscured by other layers of the build they are a godsend.

So photograph and document everything...

SAM_7711.JPG

Even the cat

SAM_7633.JPG

[bepone]

cat is suspicious!

[Stephen1966]

Perhaps someone could explain this to me?...

Here is a picture of the current limiting resistor installed prior to forming the caps....
I got a 140VDC voltage drop with this resistor, so today, I'll be forming the caps again without it, with full power.

The voltage drop didn't occur after the resistor however, but before it, at the tail of the rectifier.

all normal , you have bleeder resistor to the ground from the cap Ub+ to the GND, forming divider..

If you tie a high value resistor between the output of the rectifier and 1st capacitor and measure at the rectifier, the DC voltage will not be smoothed at this point by a reservoir cap. So your DC voltmeter will measure the effective voltage of the non smoothed DC voltage.

Both of these answers are correct. The rectifier output will have a huge ripple riding on some DC, with p-p equal to the peak rectified voltage minus the smoothed DC. In your case the lower leg of the divider is about 150k, basically all the balance resistor series pairs connected in parallel. Measuring DC, about half(?) of the the voltage drop from the peak rectifier output to the smoothed DC voltage occurs due to the ripple.

The start up procedures in circulation all seem to point to installing all the preamp tubes first and taking their voltage reading for confirmation, then it seems to be a good idea to turn the bias pot all the way down to its coolest before then installing the output tubes. Is this the time to remove the PI?

I don't know why people like to start with the preamp tubes. I prefer to get the power amp running first, which brings the B+ voltage down near normal. Of course you need to connect a load whenever you power up with output tubes installed. I check the output transformer before installing it to confirm its turns ratio and phasing, so there is little or no chance of oscillation due to positive feedback. Some manufacturers show phase dots on the schematic, but there is always the possibility that the lead wire colors are reversed so I check anyway. Breaking the NFB loop or removing the PI tube allows getting the power tube bias in the right range first.

If you get it wrong and it squeals, swapping the phases at either the pi output or at the output transformer primary leads, whichever is easier, will fix it. PI plate loads stay as they are.

Thanks Martin, you seem to have a gift for explaining these kind of things so well.

Why preamp tubes first? Maybe because they pull so little current but after this, I'm ready to try it your way with output tubes first. It will give me an earlier indicator of the adjustments I need to make to the dropping resistors in the PS. The cap forming process is also surely a stress test on all the components in circuit, so after forming I'm confident it will be robust enough to handle the demands of the output tubes. I can see the advantages of the pulling the PI as suggested as well. It's a shame, I didn't check the OT before installation, given my earlier drama with the PT I was thinking I would like to have done some testing on the PT too.

Swapping the OT primaries or the PI output - excellent. Relief I don't have to mess with the plate loads.

[Stephen1966]

cat is suspicious!

She should be

[rootz]

What a beauty this is turning out to be Stephen. Very clean work and I love a lot of your innovations, like placement of all the relais or the purposely designed PSU. Well done!!

Now what can go wrong on first power up? I never had any catastrophic failures. If your bias is set to its minimum, say -50V, not much can happen at this point if you know what you're doing. Check if heaters are working, check if a load is present (like absolutely present!) and go. Keep your hand at the Stand By switch. If you hear a very very! loud fart: swap OT primaries! Downside is that if there is a interruption between the output jacks and load (load is not connected properly), you might blow your OT if you hit the 50% chance of getting your primaries the wrong way around.

What works better and is safer: disconnect GNF at the selector as suggested before. Run power up and confirm the amp makes sound (signal in effectsloop if no preamp tubes present). You hear a sound? Okay, load is connected properly, so now you can connect the GNF and see if you got the primaries right.

So in short what can go wrong:
- no negative bias voltage (check this in stand by mode);
- primaries wrong and no load (easily overcome by above mentioned suggestion.

[rootz]

Just a small suggestion for all less experienced amp builders out here: never cut your OT primary leads short before first power up. Do this only when you are sure they are connected right. You can swap PI plate leads, before or after the coupling caps, but swapping primaries is much much simpler IMHO.

[Stephen1966]

Just a small suggestion for all less experienced amp builders out here: never cut your OT primary leads short before first power up. Do this only when you are sure they are connected right. You can swap PI plate leads, before or after the coupling caps, but swapping primaries is much much simpler IMHO.

Perhaps for some of the more experienced builders as well: I notice MrD had to use a few cable crimps in the #124

And thanks for the heads up with process, it's basically what I'm thinking as well. I.e. I am thinking about quite basically compared to your understanding of it.

We can set the bias to minimum before any tubes go in just by connecting the DVM to grid pin 5 under power. Earlier on, as I was checking the PT HV lines, I also checked the bias filter and that read around -60VDC so with tubes in, I expect it will drop to around -50V (20% is the guestimate). Heaters are working - the heater supply feeds the power LED and it's giving a nice incandescent glow as I write but the taps are rated at 6.5V so I will be checking this as well.

I actually built a dummy load some time ago (and kind of hijacked Martin's thread 'Dummy Load' with the build - sorry Martin!) https://ampgarage.com/forum/viewtopic.p ... 7#p417557 I think you mean with a speaker though - right?

I hadn't thought of feeding a signal into the FX loop before installing the preamp tubes but it makes perfect sense, and disconnection of the GNB - very easy to do. Excellent idea, thanks.

Thank you also for your kind words. A coincidence perhaps, but going all sentimental and Ken Fischer, I was thinking of naming it the Black Beauty. (English kids over the age of fifty will know the connection) It wasn't all me of course, your early input was immense and instructive. And Martin's precision PSU designs are really the godfather of this one. Regarding the relays - it's tight in that end of the chassis. These TT chassis' are not as deep as the standard Dumble ODS and just a few millimeters makes a big difference. I can't tell you how many hours it took finding the right displacement but when you look at the #060, I think you can see MrD did something similar - tucking the relay under the switches. I use smaller standoffs there, too but I think he just went with a couple of washers!

What can go wrong? Well, I've been lucky so far. And if the the OT schematic is correct (I think it includes the phase dots Martin mentioned) and it all looks good so far, touch wood, nothing is going to bang!

[Stephen1966]

Here is the cap forming procedure I used.

Three times! The first two times were with the 100k resistor on the B+ supply, between the rectifier and the PSU.

The first (ultra conservative) attempt was with a light bulb limiter. Increasing the voltage on the variac in 40V increments to achieve 108.3VDC at 240V. Each cycle involved starting the variac at 0V and slowly (like 1V per second) increasing the voltage to target. Then, at the end of each 30 minute interval, dropping the voltage back to 0V on the variac and allowing the caps to discharge before starting the next cycle. At around 160V on the variac, the LBL actually started to draw more current than the amp so I got a nice dim glow from the LBL but a decrease of 23VDC when I started the next cycle at 200V.

Conculsion: the results are barely worth quantifying since with the LBL and the 100k inrush suppressant resistor acting as a voltage divider the temperature of the caps barely rose above the ambient temperature, and B+ voltage never rose above 123VDC at 160VAC on the variac. Flicking the variac switch on and off at 240VAC at the end of the process, I got a dim glow from the (60W) light bulb but hardly any brightening with the initial inrush of current.

Next cycle - without the LBL but still with the 100k resistor. This time raising the variac in 50V increments, same as before, slowly increasing the voltage from 0V for every 30 minute cycle. This time I started to measure the Vd across the resistor:

Variac (VAC) / VDC (B+) / VDC (cap)

• 50 / 72.2 / 31.09

• 100 / 142.8 / 61.38

• 150 / 218.2 / 93.8

• 200 / 245.9 / 105.92

• 240 / 311.4 / 133.8

Conclusion: again, apart from the PT which rose by about 2°C above the ambient temperature, all the caps stayed at ambient temperature and so, while the current limiting resistor may provide a measure of protection for the capacitors at initial start up and forming, this is not really allowing us to get the temperature increase in the dielectric of the caps for them to form successfully. Maybe a smaller resistor? I believe that it is a good idea to include the resistor in the first forming since if anything is going to explode, this is where it is most likely to happen. And so, cap forming could be performed in two stages: one with the resistor, one without. Neither without the use of the LBL though - it sucks too much current from the circuit. Note, I was increasing the voltage each time, according to VAC increments on the variac. After figuring out that the resistor was causing the huge voltage drop to the capacitors and why (see above) I decided that it would be safer to measure the voltage increments in the final forming from the cap i.e. from the top of the first reservoir cap in the circuit.

This time, I was able to increase the B+ voltage in 50VDC increments and read off, the equivalent VAC increments at the variac. I increased the cycle time for each increment to 1 hour and I started to be a bit cleverer with the temperature taking - checking the PT, the PSU, the can cap and the FET filter cap which is like the ninth stage of the filter circuit, on the FET board.

VDC (B+) / Variac (VAC) / temperature

• 50 / ? / ambient

• 100 / 53.5 / ambient

• 151 / 79.2 / ambient

• 201.18 / 105.5 / ambient

• 251.8 / 131.9 / 1.5°C (PSU), 2.5°C (PT)

• 301.1 / 159 / 2°C (PSU), 4.2°C (PT)

• 351.4 / 184.5 / 3.1°C (PSU), 5.6°C (PT)

• 398.7 / 209 / 3.3°C (PSU), 8.7°C (PT), 10.5°C (FET)

• 463.4 / 240 / 4.5°C (PSU), 10.9° (PT), 6°C (FET)

Conclusion: these results should be read as merely indicative rather than conclusive in any form. I would have needed a more controlled test environment, and better test equipment along with a much larger sample range for that. Since temperature increase was essential to the forming I used a thermal leak detector to take the temperature readings but it is far from precise and this probably accounts for the temperature drop of the FET filter cap at the end of the process. It didn't cool as far as I could tell. The PT exhibits the greatest temperature increase but even here, it is really nothing more than warm to the touch after 1 hour at full power and around nine hours of nearly continuous use.

Now, considering I don't have an ESR meter, I don't have any data that would indicate the useful life of the capacitors but the process as performed is probably going to provide the caps with a healthy conditioning at the start of their working life. A consequence of the forming procedure that I see, is that it also provides a stress test for all the other components in circuit, before the tubes are installed and before any further adjustment and biasing takes place. Nothing blew up, there are no obvious shorts anywhere and while temperature increases were seen in the PT and caps, the rest of the circuit remained at an ambient temperature indicating it is good to go to the next stage.

I don't have any problem with people saying none of this is necessary. I'm actually inclined to agree. My first amp didn't have any cap forming it was just flip the switch and see! And again, the world didn't stop turning. In any case, I'm only aiming for a general procedure and results, here. I don't think it is a critical step, rather, a useful one.

[bepone]

now is too good!

[Guy77]

One thing I really liked about what you did Stephen was how you brought up the voltages very slowly on the amp.
I had a situation one time were I was bringing up the voltages very carefully with my variac on an amp and I noticed a faint smell from the amp. Low and behold there was a tiny wire , like the thin pieces of wire that you get when you are stripping the ends of shielded cables for some of the grid connections on your preamp tubes to the controls on the amp. One of these tiny strands of copper wire was touching the pins of a preamp tube and causing one of the dropping resistors in the power supply to slowly start to melt!


Cheers

Guy

[Stephen1966]

One thing I really liked about what you did Stephen was how you brought up the voltages very slowly on the amp.
I had a situation one time were I was bringing up the voltages very carefully with my variac on an amp and I noticed a faint smell from the amp. Low and behold there was a tiny wire , like the thin pieces of wire that you get when you are stripping the ends of shielded cables for some of the grid connections on your preamp tubes to the controls on the amp. One of these tiny strands of copper wire was touching the pins of a preamp tube and causing one of the dropping resistors in the power supply to slowly start to melt!


Cheers

Guy

Melting a dropping resistor - that takes some effort But you're right. Sense of smell - v.important.

[Stephen1966]

On to starting up...

I adjusted my bias pot to the minimum voltage - pin 3 was around -70VDC with no tubes so I put the wiper (pin2) down to -70VDC.

I disconnnected the NFB from the impedance switch.

I connected a load - my speaker in the Tweedle Dee.

Installed the tubes.

Calibrated the variac to 240 VAC

Power on - checked the heaters (all around 6.5VAC) - good.

Standby on. Quickly check the B+ voltages in the PSU. Incredibly, or perhaps it should not have surprised me, the B+ voltages were all pretty much where they should be:

B+1 = 441.4 (440)
B+2 = 438.5 (437.4)
B+3 = 360.7 (368.9)
B+4 = 408.5 (409)
B+5 = 332.6 (336.2)
B+6 = 320.6 (323.6)
B+7 = 310 (312.5)
B+8 = 301 (303.8 )
B+9 = 300.65 (300)

*the voltages in parentheses were the simulated target voltages. This all looked pretty good to me.

Next measured the plate voltages and again, they were pretty close to spec.

V1 - 197.5 (190)
V2 - 207.12 (200)
V3 - 358.4 (365.5)
V4 - 213.9 (220.5)
V5 - 206.6 (212)
V6 - pin 1, 285.6 (290) - pin 6, 277 (280)
V7 - 438.4 (440)
V8 - 438.2 (440)
V9 - 438.2 (440)
V10 - 437.6 (440)

No other thing to do but to plug the guitar in

Early impressions... It's very quiet (hiss) dead quiet really, unless I stand too close to it with the guitar. A quick run through of the various switches, FET, PAB and OD - all work as expected. Tone controls and EQ switches are all ok. The presence isn't noticable but the tubes haven't been biased yet and I will look into what factors will bring it into play. The OD trimmer is responsive but set very roughly. The PI trimmer, set at midnight, is just left as it is for now - looking at the plate voltages, it looks pretty close. Reverb isn't in circuit yet, I didn't have a tank connected. I did run a line from the preamp out directly to the master but didn't notice any difference when I removed that line to include the reverb dry mixing resistors in the path from preamp to master.

I've left it there for tonight and hit the bottle in celebration. Many questions answered today, the least of which being was it worth it?

I would say so.