Bottlehead Forum
Bottlehead Kits => S.E.X. Kit => Topic started by: JeffYoung on September 13, 2017, 11:37:50 PM
-
Being not generally a fan of the "top plate" style of amplifier, I'm constructing some custom casework for my S.E.X.
Top and bottom are 2mm copper plate, machined and then pickled in vinegar & salt and baked for 5 min at 450º. Since the connectors & controls are on the front and back, I've gone for a square plan (250mm x 250mm), which also gives me a bit more width to fit some bigger caps under the hood.
Sides are 1/2" x 2-1/2" aluminium bar; front and back are 0.090" aluminium plate. The front, back and sides are inset about 1/16" to give a bit of an "ice-cream sandwich" look. (In hindsight I think 1/8" or 3/16" inset would have looked better.)
The holes in the back might look a little odd: I've switched out the speaker posts for some vintage-style ones to go with the casework better. Same for the front: the power switch is rotary (with a chicken-head knob), while the volume pot is an Alps with a big Dakaware knob.
Front, back, sides, output chokes and transformer bell painted in wrinkle-finish paint.
-
That'll be a cool looking amp!
-
Looking forward to seeing the finished product. Please include build photos too, always nice to see someone's vision come to life.
-
I've got a bit more done, but I didn't take any pictures yet.
I'm upgrading the interstage and parafeed caps, so I decided to make use of the little Dayton 0.1uF's by putting snubbers on the two transformer secondaries. I used Quasimodo (http://www.diyaudio.com/forums/power-supplies/243100-simple-math-transformer-snubber-using-quasimodo-test-jig.html (http://www.diyaudio.com/forums/power-supplies/243100-simple-math-transformer-snubber-using-quasimodo-test-jig.html)) to stimulate ringing in the transformer secondaries and came up with the following CRC networks to more-or-less critically dampen them:
Filament secondary
Cx: 0.01uF Illinois 220VAC polyester
Cs: 0.1uF Dayton 400VDC polypropylene
Rs: 8 - 10 ohms
B+ secondary
Cx: 0.01uF Illinois 220VAC polyester
Cs: 0.1uF Dayton 400VDC polypropylene
Rs: ~ 400 ohms
Cheers,
Jeff.
-
Heater wiring 95% complete.
-
The shielded twisted pair that we provide with the kit may offer lower noise than just a twisted pair of cables.
Also, the 0.1 Ohm 5W resistor provided is not in the signal path and does not benefit from being replaced with a different part.
-
Hi Paul,
The bulk of my changes are just about looks. I like the look of the Ohmites, so I'll also be using them in the CRC filter for the B+ supply. 8)
Same for the twisted pair heater wiring. I figured I could get away with it on a DC heater, and the twisted pair is easier to keep at 90º to the signal components (which will hopefully mitigate any noise issues).
Another set of changes is just to match my "house" style: I'll be separating the signal ground from the chassis ground, and I'm using the same wire colours as in my other builds.
I am making a few changes for sound: larger caps in the CRC filter (with the last stage bypassed by film caps) and upgraded interstage and parafeed caps.
Cheers,
Jeff.
PS: I don't kid myself that the transformer snubbers are likely to improve the sound either. But they might, and they were fun to research and implement.
-
I'll be separating the signal ground from the chassis ground,
They need to be connected somewhere, as well as the power supply ground.
-
They need to be connected somewhere, as well as the power supply ground.
Yep; no worries. The chassis will be connected directly to the IEC ground, while the signal ground will be connected via opposing 600V 35A diodes and a 680V 8A thermistor.
Cheers,
Jeff.
-
the signal ground will be connected via opposing 600V 35A diodes and a 680V 8A thermistor.
I would strongly recommend not doing this.
-
Hi Paul,
The same discussion was had over on DIYAudio forums, but no one ever gave a good reason why (other than that a wire was "safer").
If anything over 1.1V leaks to the signal ground then those diodes are going to look like a wire plenty long enough to vaporise the IEC fuse (or trip the residual-current-detector). And if there should be some failure mode that presents less than 1.1V with high current, that thermistor won't even be warm before the fuse meets the same fate (or the RCD goes).
I can understand if the failure mode of diodes + thermistor has more to do with lawyers than electricity, in which case no more needs be said.
But if there's some other failure mode of the diodes + thermistor approach I'd love to hear it.
Thanks,
Jeff.
-
Yes, the wire is safer.
You're adding parts to a proven design to solve a problem that you do not have, that's the core issue.
-
Fair enough. Thanks for the extra info, Paul. Your guys responsiveness is first-class.
Cheers,
Jeff.
-
At the risk of giving the Pauls a heart-attack, here's another picture of my S.E.X. developing.
Note that this build uses non-standard wiring and component positions. Don't follow this as a guide.
You can see several changes here (after all, the middle initial is for "experimenters"):
- I've added CRC snubbers to the B+ supply. (http://www.hagtech.com/pdf/snubber.pdf)
- I've changed the filament supply filter from an RC to a CRC.
- More of those Ohmite resistors.
- A Mundorf EVO al/oil for the interstage cap.
-
At the risk of giving the Pauls a heart-attack
I've changed the filament supply filter from an RC to a CRC
(https://forum.bottlehead.com/proxy.php?request=http%3A%2F%2Fstatic-14.sinclairstoryline.com%2Fresources%2Fmedia%2Fe4db02b9-eec5-451a-859c-d9bae0aebedc-large16x9_transformer.jpg%3F1470512704970&hash=539feeff894e7814c2071ead5f2a621db93717f9)
-
Nichicon PWs: 3.8A ripple. No Kaboom. Hopefully. ;D
-
The photo isn't of a cap blowing up ;)
-
You think the inrush is going to over-tax the transformer?
-
You can simulate the AC current that the transformer sees with the resistor input filter and the cap input filter in PSUD.
In one scenario, the transformer runs within its specifications for temperature rise. In the other situation, the transformer runs over its specifications for temperature rise.
Inrush current is very brief and not generally concerning, especially with solid state rectifiers.
-
Yeah, that's why I was building a model of it in PSUD2. The numbers don't change much: peak current goes from 6.4A to 7.1A, while RMS goes from 2.2A to 2.3A.
The temp rise in a transformer is going to track RMS, rather than peak, right?
-
Here are the results I get.
-
Oops, I was monitoring the bridge, not the transformer.
I still get different results, though: 3.1Arms for RC and 3.2Arms for CRC.
The primary difference is that I have the transformer at 223mΩ, whereas you are using 99mΩ.
I measured my primary at 33.3Ω and my secondary at 0.2Ω, which PSUD2 estimates as an effective impedance of 223mΩ. Is yours really as low as 99mΩ?
-
No load voltage is 6.42V (the photo in the manual has exactly 120V presented to the primary ;) ), the rating for the winding is 6.3V/3.5A. I vastly prefer to use these numbers compared to attempting to measure the DCR of a winding with such low DCR. Incidentally, the source impedance actually looks to be a little lower than I posted in that image (I was using the off load voltage from memory, probably from a different transformer).
I also noticed you went up to 15000uF caps, which drops their impedance a bit also. My load calculation on the transformer with your CRC filter is about 4.3A, with the stock configuration clocking in at 3.6A. (Do note that I have the unfair advantage of having talked to PJ about this in the past)
I've roasted a few power transformers, including one PT-7 (old SEX transformer) last week, and it's certainly a stinky experience!
-
Yep, the transformer numbers definitely make a difference. Using my no-load voltage (6.64V) and the spec'd power rating gives me 97mΩ, which in turn yields 3.4A stock and 3.7A w/CRC.
The remaining difference is probably down to the rectifiers we're using in the model. Your 1N5828s have a lower voltage drop than the MBR745s used in the kit.
Here's a reasonably accurate model of the MBR745s if you want to dump it in your PSUD2 rectifiers.txt file:
MBR745, SS, 0.007, 9, 4000, 45, 150, 7.5
In any case, sounds like I need to keep an eye on transformer temps. I assume it will get externally warm to the touch before it overheats?
-
I calculated the ESRs of the two capacitors based on the formula tanδ = 2πfCR and got 20mΩ and 38mΩ at 100Hz. (Not trusting my math I also looked up a bunch of 15,000uF/10V capacitors that did publish ESR specs and they were in the same ballpark.)
Those ESRs (in my model) produce 3.6A CRC and 3.3A stock.
-
Oops, I was off by a decimal on the cap impedance (at 120Hz for where I live).
Still the CRC is ill advised.
-
I took some temperature readings running the glow test for 30 min. I used the choke readings as "ambient" to remove any emissivity errors (they're painted with the same black wrinkle-finish paint as the transformer bell).
Temps increased for roughly the first 8 minutes, and were then stable for the next 22. Filament voltage was stable at 6.6V.
Choke: ambient
PT-10: + 3ºC
0R1: + 19ºC
C-ripple: + 3ºC
C-smooth: + 2ºC
There was of course limited drain on the B+ side, so the temps might come up a bit with the whole amp running. The filament voltage will also likely drop a volt or two, which might mitigate the rise to some extent.
Paul's advice is noted, but these seem pretty benign so far....
-
Without a load on the HV winding, it will offer a nice, cool layer of copper to sink heat into.
I can offer that when we (mostly PJ, I was just an interested observer) wanted to measure temperature rise in a transformer, we would let it run for 24 hours under its intended load. We (we is again referring to PJ) also would use the change in resistance of a given winding to calculate what the actual internal temperature of the power transformer was at that time with a meter intended to provide accurate DCR measurements at very low values.
There's a huge difference between the temperature of the copper inside the transformer and the temperature of the lamination stack wrapped around it.
-
Thanks for the info, Paul.
I'll do a longer test once I get the B+ side under load.
-
Chassis:
-
It's coming along nicely. Just about ready for my resistance tests.
(https://forum.bottlehead.com/proxy.php?request=http%3A%2F%2Foi63.tinypic.com%2F2vx3wg3.jpg&hash=f3de2880771ad18d749be8e21f990c7e4776c244)
-
If you left off the 8.2K resistors (it sorta looks like they aren't in there), then the amp will work but may be a little buzzy. This may go away with installation of the C4S upgrade that I see in the background.
-
Hi Paul,
I was originally trying to save room on the top lugs for film bypasses on the 22uF caps, so I mounted the 8K2 resistors in-line at the bottom:
(https://forum.bottlehead.com/proxy.php?request=http%3A%2F%2Foi65.tinypic.com%2F23lml2o.jpg&hash=fa7d0f1727ae12c7abd45e12addaf353d6527bcc)
Later I concluded that the C4S's would make bypass caps unnecessary so I left them out.
I put the CS resistors on the ends of the flying leads for the C4S boards for now, as my case design will make it difficult to get to them after-the-fact. (I now fully understand the appeal of everything-on-the-top-plate design. ;))
(https://forum.bottlehead.com/proxy.php?request=http%3A%2F%2Foi66.tinypic.com%2F2nsncwg.jpg&hash=589c1ce171e6708b7a43ceae956c5152dc467afd)
Cheers,
Jeff.
-
Ah, nice, you tucked those away nicely.
The top plate design sure does make things easier to build.
A 22uF/450V axial cap will also fit nicely and sit between the two T-strips if you want to complete the look a little more ;)
-
Finished base kit (haven't installed C4S yet):
(https://forum.bottlehead.com/proxy.php?request=http%3A%2F%2Foi68.tinypic.com%2F9pltzs.jpg&hash=8c00525d09dc82ac5e0b8960da06285de64c3df6)
(https://forum.bottlehead.com/proxy.php?request=http%3A%2F%2Fi66.tinypic.com%2F2vt16oh.jpg&hash=6065f2efcc1d2ca1230e5d2e0b35fd56e26ab55c)
-
That looks classy! Love it.
If I may suggest something, I feel the knobs are an eye sore. Metal knobs would add to the class factor and fit in with the overall design. Just my opinion. Still looks very nice.
-
I think the Dakaware knobs are cool.