Post by sadface on Jul 14, 2020 17:53:51 GMT 12
G'day Guys,
I've been planning this one for quite some time since I got fed up with battling my SP14 from www.tubesforHifi.com
Here is it in it's semi completed original form back in 2011
A few rounds of upgrades and re-layouts have occurred in the intervening decade but I've never been happy with the layout due to the massive board being so difficult to work around within the Ezchassis from Simon at Designbuildlisten. It was also built fairly early in my DIY Hifi journey so mistakes were made and I have learned much since then.
Thus it is time to start again with a clean sheet design and go truly dual mono (the claim was made to dual mono, but it merely has independent regulators for the b+ and heater supplies for each channel off shared transformers)
I looked hard at the dual mono boards from Tubecad.com but Broskie's Boards are a bit large for my purpose and I don't quite like the way he has arranged wiring connections. What I really want to do is mount the signal stage as close to the back of the chassis as possible. The stepped attenuator and the source selector switch will require shaft extension for the same purpose so that the wire runs can be kept as short as possible.
With this in mind I have decided to lay out my own boards.
B+ will be regulated at 265v. HV regulator boards are coming from Linear Audio. They might have left Belgium by now, who knows with current international shipping. I am yet to design boards for the heater supplies but I have opted for 12.6vdc regulated with each pair of tubes wired in series to lower the current requirements.
I have ordered a pair of transformers from Brian Smith in Foxton. 230v @60ma and 12.6v @ 2A. I decided to oversize the transformers for a nice cool operation and hopefully long life.
Here is my current pcb layout for the signal stage. Top side
Bottom side
Space has been allowed for some truly massive coupling caps.
Footprints are present for floating the heater supply at 1/4 of b+ on board. Faston connections for power. Screw terminals for signal wiring. The ground plane has been separated into 3 sections which meet at the ground faston, signal, HV decoupling and heater floating.
Any critiques? Anything wrong or patently braindead?
|
Post by Owen Y on Jul 14, 2020 18:30:27 GMT 12
Are the coupling caps for C3 (output) position?
|
Post by sadface on Jul 14, 2020 20:13:37 GMT 12
Hi Owen,
Yes. I based the schematic off the SP14 schematic.
I've been too lazy to change it to C1.....
|
Post by colinf on Jul 18, 2020 18:54:07 GMT 12
Looks ok, the only thing I’d do to make it easier to solder would be to use thermal relief pads for each of the connections to the ground plane on the bottom. Dual mono grounding and boards is good! ...‘Battling my SP14’, how?
AMR-iFi R&D
|
Post by sadface on Jul 19, 2020 17:01:29 GMT 12
Looks ok, the only thing I’d do to make it easier to solder would be to use thermal relief pads for each of the connections to the ground plane on the bottom. Dual mono grounding and boards is good! ...‘Battling my SP14’, how? Hi Colin,
What do you mean by thermal relief pads? Are you meaning something like extra through holes to allow air flow?
The battle was referring to getting a layout that satisfied me. Things like minimising wire runs, keeping the transformers as far away as possible etc. Because the board is so large I could never use shaft extenders where I would like to.
|
Post by colinf on Jul 19, 2020 20:14:17 GMT 12
Oh I see, SP14 from tubes for Hifi. (The above link doesn’t work.) I thought you meant Audio Research SP14 line stage. Thermal relief pads are component pads designed for connection to ground planes. A normal pad has the ground plane connected straight to the pad all around it’s circumference. When you come to solder it it’s quite hard as the soldering iron has to heat up the surrounding area of ground plane, which takes a lot of heat and time from the iron. A thermal relief pad has small spokes, usually 4 spokes, around the pad that connect it to the ground plane. That way the pad can be heated efficiently while still being connected to the copper plane.
AMR-iFi R&D
|
Post by sadface on Jul 19, 2020 20:48:16 GMT 12
Hi Colin,
Thanks for explaining that, I've been wondering what the deal was with those pads.
I will have to figure out if Easyeda can make them.
|
Post by sadface on Oct 28, 2020 20:49:28 GMT 12
G'day Guys,
A wee update on this build.
Some parts have arrived from both Belgium and Foxton.
Firstly a pair of lovely custom power transformers from Brian Smith in Foxton.
12.6v @ 2A 230v @ 60mA
Secondly the high voltage power supply boards arrived from Linear Audio in Belgium.
|
Post by sadface on Dec 2, 2020 19:40:49 GMT 12
G'day Boys,
With the completion of my last gainclone project I can now properly start on this one.
The first step is to get rid of the old board.
My plan here is to remove everything or as much as possible, clean everything up and sell it all off with the board, original documentation and parts kit.
I think it is best to take it back to bare board etc so that nobody has to deal with the mistakes I made 10 years ago....
If anybody is interested, flick me a message and we can sort something out. Much easier to go local than offering it up overseas
|
Post by Owen Y on Dec 2, 2020 21:47:24 GMT 12
My feelings about 'dual mono' have changed over the years - I think now that it's more important to 'separate' your AC PSU parts from your 'signal circuit' parts - rather than rigidly obsess about 'dual mono' (for stereo channel separation reasons or whatever). Especially true for low signal ccts like preamps & esp. for high voltage, high current tube ccts. How did you decide to reconfigure this preamp - in principle?
|
Post by sadface on Dec 3, 2020 19:57:43 GMT 12
The core issue was that the giant monolithic pcb was to difficult to work around within my desired chassis layout.
I want to minimise the wire runs by running extension shafts and mounting the volume pot and source selector switch right at the back of the amp. With the existing board there are tubes in the way of where the shafts need to go.
By separating things out into 4 boards per channel I should be able to put things exactly where I would like them: Signal board, b+ reg board, b+ rectifier board and heater reg board.
Some other advantages: The T-Reg boards I am using for B+ are quieter than the LR8 regulator used on the existing board. It also has extra built in protection.
More capacitance after the B+ rectifier due to solid state diodes.
Better on board filtering on the signal boards. One less power transformer compared to the old setup.
I am considering adding an extra steel panel between the power transformers and the rest of the amp for extra shielding.
Here is a concept diagram.
|
Post by Owen Y on Dec 3, 2020 21:25:05 GMT 12
Layout looks good, always good to see low-level signal runs shortened with sel. switching & pot near inputs. Yes, can usefully put a steel shield up across the PTXs.
|
Post by sadface on Dec 6, 2020 8:50:16 GMT 12
G'day Guys, I'm starting work on the rectifier board for the B+. I am debating whether it is worth putting and RC filter on the rectifier board before b+ regulator. I have some Epcos 220uF 450v caps for the purpose. The simple approach would of course be to simply chuck a cap after the rectifiers and leave it at that. The slightly more costly approach would be to chuck something like a 100R resistor before a single reservoir or nest 100R in between 2x 220uF caps for an CRC filter. Both would have a corner frequency of 7.23Hz: The RC filter would be down 17dB @ 50Hz and 23dB @ 100Hz. The CRC filter would be down 34dB @ 50Hz and down 45dB @ 100Hz. By my calculations, the preamp would be drawing circa 20mA per channel which equates to a 2v drop over a 100R
The power transformer is 230v for the B+. I am using HER208G rectifiers which have a forward voltage drop circa 1v @ 20mA. Target B+ is 265v at the output of the regulator. 230v * 1.41 = 324V at the rectifier. 324v - 1v - 2v = 321v. Call it 320v at the input of the regulator which should be plenty of headroom for regulation.
So the question becomes: Is it worth the extra cost, complexity and voltage drop to add a passive pre-filter before the regulator.
|
Post by colinf on Dec 6, 2020 21:37:20 GMT 12
|
Post by sadface on Dec 7, 2020 16:19:31 GMT 12
Does a corner frequency of 7Hz sound about right?
220uF - 100R - 220uF is what I am thinking of right now.
|
Post by Owen Y on Dec 7, 2020 17:55:11 GMT 12
- I know that Morgan Jones says that RC filters are 'cumulative' through a cct & that you should therefore design for something as low as 1Hz - but that's sometimes difficult to achieve in a PSU filter. He also says that as f3 'corner' freqs are -3dB, the 0dB roll-off point of the filter is actually around 7x the f3 freq. ie. If you want it flat response down to say 20Hz, then you should be designing for an f3 of 20/7 = ~3Hz. - I also like to use PSUD2 (Power Designer 2) to model PSUs. To check ripple, step (transient) response, etc.
|
Post by sadface on Dec 7, 2020 18:45:31 GMT 12
|
Post by Owen Y on Dec 7, 2020 19:05:20 GMT 12
PSUD will tell you if you have any turn-on peaking or not. Yes I think that large capacitance can tend to sound 'slower' (which is why some use paralleled smaller caps). However others say the 'time constant' of each filer 'cell' is key, eg. in a preamp, keep the TC of the 1st PSU RC cell less that 15mS. But in your case you have a regulator inserted, so maybe all this is moot
|
Post by sadface on Dec 7, 2020 19:58:13 GMT 12
Here is my proposed heater regulator:
Here is my proposed B+ rectifier board
This would be the complete B+ supply
|
Post by sadface on Dec 9, 2020 20:25:39 GMT 12
G'day Folks,
Work continues.
Here is the first draft of the heater supply pcb
I considered going with something more exotic than a simple LT1086 but I decided that a heater doesn't need a super low noise supply.
I think the LT1086 should be sufficiently quiet.....
Meanwhile I am butting my head against the brick wall of physical space inside my chassis. It is becoming apparent from mock ups that I probably won't have sufficient space to lay out how I would like to within the chassis. This puts me in a serious dilemma as I have a 2nd chassis that is supposed to get a DAC installed in it soon (more on that later). I also have a matching plywood front that fits quite nicely on my Marchand crossover.
The idea was to have some lovely matched chassis for preamp, crossover and DAC.
It might well be that I have to change tacks and use some the various salvaged home theatre amp chassis I have been collecting recently: Give them all matching wooden front panels and perhaps strip all the paint off and replace with something cool like Rustoleum.
First World Problems......
|
Post by sadface on Dec 12, 2020 21:47:56 GMT 12
G'day Guys,
Here is my first draft of the B+ filter board I didn't do my usual pours for the traces on this one as we're talking 20mA.
This board is a bit larger than I would like it to be. I wanted to increase the spacing as much as practical for the purpose to clearance to prevent arc over.
There is at least 5mm between between pads of opposite polarities. I might still consider using nylon screws and standoffs for extra safety.
I am considering whether I might want to add a bleeder resistor however I think the regulator afterwards should fairly quickly drain reservoir caps. My thinking is the same for the heater regulators, the heaters themselves should quickly drain the reservoirs.
Regarding the chassis situation:
Assuming I can't make everything fit nicely once the boards arrive into my existing chassis, I have a Denon and an Onkyo chassis that I think I can make match nicely. The home theatre amp chassis are a good 50mm wider internally which should provide the extra space to get the 'perfect' layout.
I am thinking that hammered dark bronze rustoleum would go quite nicely with a stained brown wood front panel like on my last gainclone project.
|
Post by Owen Y on Dec 13, 2020 21:23:18 GMT 12
I mostly don't use bleeder in my valve ccts, even power amps, never seem to need them
|
Post by sadface on Dec 17, 2020 19:54:50 GMT 12
G'day Guys,
Progress is progressing.
I've done a fair bit of ordering recently: Mouser, Rs components, Element 14, JLCpcb and even Ebay. I should have all of the electronic components required before christmas (hopefully).
I ordered some tube sockets from ebay last Tuesday and they miraculously arrived this Tuesday despite the economy shipping on China Post. Expected delivery date according to Ebay was mid feb....
Unfortunately the ebay tube sockets turned up an annoying error. I didn't properly check out the footprint I used for the tube sockets on the signal stage board design. I have 1.27mm diameter holes and tube sockets with 2mm wide pins..... More annoying, I paid the extra $20 for ENIG finish on the signal boards which are probably now useless..... Since the pins on the tube socket are flat, I might be able to bend them into a U shape and squeeze them in. I will wait to see what arrives before ordering updated boards.
I also ordered some spare boards for the LM1875 and some other projects I have in the pipeline. I decided to try some different colours so I got some in white, yellow and red boards to see what they come out like.
I also ordered a relay input selection switch kit from Jims Audio (via Ebay) as one possibility for solving some of my space problems with the existing chassis. Once everything arrives I can do some more accurate mock ups and see what may work.
|
Post by Owen Y on Dec 18, 2020 22:06:54 GMT 12
Maybe use some PCB stakes in the socket holes, then drop the socket tags over the stakes. (If you've got PCB-pin sockets, you could swap to traditional tag-base sockets.)
|
Post by colinf on Dec 19, 2020 20:37:38 GMT 12
Perhaps snip a little bit off each socket pin. Otherwise keep them for a later project on which you use the right hole size on the board, and order some different sockets for this project.
AMR-iFi R&D
|
Post by sadface on Dec 23, 2020 16:39:57 GMT 12
G'day Guys
Thanks for the ideas.
I have ordered another set of sockets. These ones look likely.
My pcbs and almost all of the other parts arrived in the last day or 2. The ENIG finish of the signal stage pcbs looks pretty nice.
The tube sockets I have most definitely do not fit the hole pattern or hole size. I could conceivably trim them down and bend them inwards to make them fit but I will see how the next set of sockets turns out before applying such drastic measures.
While waiting for the next batch of tube sockets, I might start populating the PSU boards. Probably more likely, I will do some work on some other projects.....
|
Post by sadface on Jan 24, 2021 15:56:46 GMT 12
G'day Guys,
The second set of tube sockets I tried were not going to work either.
So I finally bit the bullet and got to work on the octal socket footprint.
I edited the footprint to have appropriate hole spacing to fit the pcb mount octal sockets I have which have a 25mm pitch between opposite pins and 10mm pitch between adjacent pins.
I figured out how to make thermal relief pads in EasyEDA.
Here is my new version.
I am considering removing the separation between the 3 grounds. Right now, ground is divided into a signal ground, b+ supply ground and heater supply ground. The 3 connect up at the B- connection. If I was to remove the separated grounds I could also flood the top side with a ground plain as well.
Thoughts?
|
Post by colinf on Jan 25, 2021 20:51:42 GMT 12
Good idea, make the signal ground path as short as you can make it. Also take into account that the power supply is in series with the signal. Ground planes have been around a long time and are known to work well. But you need to be conscious of the current paths in a ground plane as well. Electricity takes the path of least resistance and a small, concentrated cluster of ground connections can have even lower resistance (and voltage gradient across it) than a ground plane.
AMR-iFi R&D
|
Post by sadface on Feb 1, 2021 7:39:18 GMT 12
G'day
Here's the updated version with top side flooded with ground.
|
Post by colinf on Feb 1, 2021 20:36:09 GMT 12
Nice. Where you’ve got B+ and B-, it should be B+ and Gnd to disambiguate the B- label. B- usually means a negative voltage rail.
AMR-iFi R&D
|