Post by sadface on Oct 30, 2018 16:40:02 GMT 12
I will also ask the question here. The preamp board has no obvious point for a ground to chassis, that is to say none labelled.
In its previous life I simply left it ungrounded as might be seen in the previous photos.
The chassis is earthed to the IEC socket. The transformer is presumably grounded to the chassis via its chassis, but nothing going out of the preamp pcb other than the I/O wires. The RCA sockets are insulated types and once again only grounded to the pcb.
My guess is that this would be nothing like best practice?
In terms of fixing this.
I thought about grounding the preamp pcb to the amplifier boards but I could see this creating a ground loop between the 3 boards and the main star. Possibly only ground to one amp board?
Seems like a better way would be to ground the preamp pcb directly to the main star, perhaps off the power supply end of the pcb so its downstream of the signal. I'd then run screen wires along the signal wire to the power amp boards but only ground it at the preamp end.
|
Post by Owen Y on Oct 30, 2018 20:06:26 GMT 12
Common good practice is to connect 'circuit Ground' to a chassis 'Earth' point near the input socket. This chassis Earth is directly connected to the mains Earth. It is also common practice to 'lift' the Circuit Ground above the Chassis Earth point via say a 10 ohm resistor (reducing ingress of Earth noise/currents, but still meeting CE safety rules) - the R often bypassed with a 0.1uF cap (allows RF to still go to Earth). This 10R is sometimes referred to as a 'Ground loop breaker' - reducing circulating currents, eg hum. (That's my amateur understanding.)
|
Post by colinf on Oct 30, 2018 20:14:05 GMT 12
Heater float like this: Ground the preamp board ground to the chassis ground (your star earth) using a 10 ohm 5W resistor. That way if you connect a source device that is grounded to the preamp it won’t result in a hum loop. The poweramp board and power supply both then derive their grounds from the preamp board. At the RCA sockets' ground use a 0.1uf ceramic cap straight to the chassis close by. That way any RF entering the unit via the connected interconnect shield gets routed straight to the surrounding chassis.
AMR-iFi R&D
|
Post by sadface on Nov 3, 2018 10:23:53 GMT 12
Hi guys, Sounds like this is what you are proposing
My original plan was to run 2 separate ground loop breaker circuits connected to the main star. One for the preamp and one for the power amp sections. I think your suggestion might prove better as well as being simpler.
Regarding the ground loop breaker. This is what I got out of the Rod Elliot's website. He recommends shorting the resistor and cap across a high current rectifier in case of fault current. sound.whsites.net/earthing.htm.
|
Post by colinf on Nov 4, 2018 5:06:22 GMT 12
That works even better. If you’re using that circuit you can get away with a smaller wattage resistor for the 10 ohms.
AMR-iFi R&D
|
Post by sadface on Nov 24, 2019 13:28:42 GMT 12
Alright boys,
After a year of inaction this build is moving forwards again.
The prototype has proven to be unreasonably good so there hasn't been much urgency. However the Christmas holidays are just around the corner so I'm preparing to keep myself busy building over the holidays.
Among the many changes to the Evil Plan over the last year has been a change in amplifiers, power supplies and enclosure designs. We're going a bit more upmarket than originally intended but that is half the fun of such a project.
Firstly I have some induced hum on the prototype due to mains wiring and transformers being to close to the speaker magnets. With this in mind the plan is now to mount all electronics within a centre steel box for magnetic and noise shielding. The speakers will be mounted in separate boxes on either side to make a single monolithic unit.
The amplifier stage will now use the LM1875 chip instead of the LM3886. Since the CHR70 speaker drivers are nominal 20w drivers I have taken the opportunity to use lower power but better sounding LM1875 chip.
The schematic is based off the old Silicon Chip Schoolies amp. The ground lift resistor has been removed, filter caps increased in size and the Bipolar electrolytic in the feedback path has been changed to a film type because I can. I originally purchased some LM1875 pcbs from Ebay however they are a bit cramped and have no fuse rails on board. So I have been designed some pcbs that will hopefully do what I want them to do. Once again these are based on the old schoolies design but the boards are smaller and have exactly what I want on them, including space for a screw driver to get through the filter caps to the chip for heatsink mounting, spade connections and space for a giant 22uf Film cap.
Here's the original schoolies PCB for reference. 80mm x 63.5mm I've tried to preserve the best features of the layout such as the separated grounds for input output and power.
The power supply is now going to use the AMB Labs sigma22 PSU as I have a Chinese copy pcb lying around. www.amb.org/audio/sigma22/
I will be wiring the psu for +/- 22Vdc as I would like to limit the output on the LM1875 chips to keep them to 20w.
Relay Speaker protection will also feature but is not yet finalised.
|
Post by sadface on Nov 24, 2019 13:54:04 GMT 12
And now the tube bits:
I have stuck with the idea of reusing the old guanzo preamp as I think it will be perfect for the application. However I've never been happy with the unknown quality of components on the board and since I was planning to redo the power supply section, I have since decided to redo the whole thing and simply preserve the signal stage schematic on a new board. Only the power transformer and the old stepped attenuator will remain.
Here is the signal stage schematic.
Since the psu will be off board I have added decoupling for both the b+ and heater on the signal stage board. Otherwise the main goal was to allow the use of larger coupling capacitors and have everything laid out to minimise wire runs when it all goes together.
On this board I have chosen to go with a solid poured ground plane on the bottom side as it seems unnecessary to separate grounds on such a simple circuit.
And now we get back to the preamp PSU. I have stuck with Colin's Mosfet design from earlier in the thread.
The odd screw holes are so that this psu board can be stacked with spacers on top of the power amp psu board.
|
Post by Owen Y on Nov 24, 2019 14:20:08 GMT 12
Looking forward to the fun part - the soldering
|
Post by colinf on Nov 24, 2019 20:02:19 GMT 12
Wow looking good Sadface! Love the regulated power supply for the LM1875, that’s unusual. Good idea to use a big film cap in place of the bipolar electrolytic. Since you’ve gone to town on the regulated power supply perhaps you could use a dc servo circuit to keep dc under control, so you can eliminate the feedback cap altogether and get even lower distortion. But the circuit board is done now. Have you been using schematic editors and pcb software for a while? Eagle? If you want regulation on the valve HT supply instead of just mosfet-assisted filtering you can replace R3 and R4 with zener diodes of the B+ voltage you want. Also the LM1875 has its own over-current protection, so you might not need an output relay. You could always have a switch-on mute circuit in between the preamp and poweramp.
AMR-iFi R&D
|
Post by sadface on Nov 24, 2019 21:58:33 GMT 12
Wow looking good Sadface! Love the regulated power supply for the LM1875, that’s unusual. Good idea to use a big film cap in place of the bipolar electrolytic. Since you’ve gone to town on the regulated power supply perhaps you could use a dc servo circuit to keep dc under control, so you can eliminate the feedback cap altogether and get even lower distortion. But the circuit board is done now. Have you been using schematic editors and pcb software for a while? Eagle? If you want regulation on the valve HT supply instead of just mosfet-assisted filtering you can replace R3 and R4 with zener diodes of the B+ voltage you want. Also the LM1875 has its own over-current protection, so you might not need an output relay. You could always have a switch-on mute circuit in between the preamp and poweramp. Thanks,
I can't say I understand dc servos at this stage so I will probably skip that idea for now. Otherwise it will be another 6 month research rabbit hole.
I've been playing with pcb design a couple of months now. I'm using easyEDA as its free. I discovered that one can order 5 pcbs for all of $3 + shipping @ JLCPCB. My first pcb attempt was for the dual rail LM338 design which featured earlier in the thread. It was ultimately a bad pcb layout but it was a great learning experience. It is interesting seeing what you design in the software versus what turns up on the courier. I learned to pay more attention to pad sizes and clearances. The most glaring issue was a lack of clearance between screw holes and traces. I could probably use the boards with a nylon washer to insulate the screws from the traces but I think I will just chuck the boards out and call it a cheap course in pcb design.
I considered going for a proper regulated supply for B+. My big tube preamp has independent regulated B+ supplies for each channel using an LR8 and TIP50, I considered pinching the design but I decided it was unnecessary. Also I would then be forced to fiddle around with the bias point on the preamp which is another rabbit hole to avoid. That's not to say it won't happen later as I will have spare boards to play with if I ever feel inclined.
The +/- 22vdc for the LM1875 is more to limit the current through the power supply. I need to drop around 9V over each mosfet to maintain regulation and prevent sag under load. Under a heavy load of say 2.5A I will need to dissipate 25W at the PSU. I do have some 63.5mm high heatsinks which will help but I am slightly concerned about heat build up in what will end up being a cramped box by the time I cram: 2x power tranny, 2x power supplies, 2x amp modules, 1x preamp pcb, speaker protection, and probably a Bluetooth receiver (and another PSU). Dimensions at this stage will be something like 175mm wide by 300mm high by 225mm deep. I am guessing that under normal usage power output will be 1W but I'd rather plan for worst case usage now than deal with it later.
Mainly I want this for muting at power on and off as apparently the LM1875 is liable to have some turn off thump. I also have a cheap ebay speaker protection board but it is a bit suspect due to its source.
The other thing of note is that I won't be moving the Volume pot to the front of the preamp like had been previously suggested. Something like if it ain't broke don't fix it.
|
Post by sadface on Nov 24, 2019 22:01:16 GMT 12
One quick question regarding pin 5 on the 6n3 tubes(2C51, 5670). Pin 5 is the shield. Am I correct that the shield goes to ground?
|
Post by Owen Y on Nov 25, 2019 8:33:40 GMT 12
Yes the internal shield can be Grounded - but not too important I believe, at audio freqs - more for VHF.
|
Post by colinf on Nov 25, 2019 9:02:35 GMT 12
Yes pin 5 goes to ground. Are you mounting the 4 power supply mosfets and the two LM1875s on the one heatsink? Under normal operation it should be fine but if you find it gets too warm you could use a bigger heatsink or use a fan.
AMR-iFi R&D
|
Post by sadface on Nov 27, 2019 21:14:34 GMT 12
The PSU pcb has space for large footprint pcb mount heatsinks. I've purchased some 2.5" tall heatsinks. 2.6 C/w and each mosfet gets its own heatsink. nz.element14.com/aavid-thermalloy/530002b02500g/heat-sink-2-6k-w-to-220/dp/2295719They might get a bit hot if used full power for a long stretch but that seems like an unlikely use case. The 2 lm1875 chips will be mounted on single heatsink from jaycar. Supposedly 0.72 C/W however I suspect that is a fan forced spec as it seems just a tad small.
No need for one of those photos other than KT88s, I'm sure you understand.
|
Post by colinf on Nov 27, 2019 23:57:49 GMT 12
The Jaycar heatsink mightn’t even get warm! Well, a touch warm but the 1875s won’t be making too much heat if the speakers are efficient. The temperature rise above ambient would be about 45c at maximum continuous power and ok ventilation.
AMR-iFi R&D
|
Post by sadface on Nov 28, 2019 20:29:54 GMT 12
Well, I've done an order of PCBs from JLC PCB. 10- dual rail LM338 +/- 30v boards 10- Dual full wave rectifier boards to stack onto the LM338 boards 10- LM1875 boards 5 - 6N3 preamp boards
5 - 6N3 PSU boards
10 - Mono NE5532/ LM19720 boards 5 - Ultra low noise LM317/337 +/-15VDC boards 5 - 15v speaker protection boards. Let me know if anybody is interested as I will certainly have spares. I won't release the speaker protection boards as they have a bit of borrowed IP which I'm not comfortable distributing. All the rest are open source schematics with my layout or my IP so I have no qualms.
I will also release the gerber files if people want to order their own.
Production is done on everything but the LM338 boards as they were 2 oz copper so they take a bit longer. DHL should get them to me by next Friday if I am lucky. Last time they took all of 15 hours to get from Shenzhen to Auckland and then 4 days to get from the airport to Wairau Park.
|
Post by Owen Y on Nov 29, 2019 11:13:36 GMT 12
Thanks, good to know about JLCPCB. I have this idea about getting black solder mask boards with white silkscreen lettering, as a way of doing in/out/switch socket front or rear panels for amps etc. Instead of phenolic or metal, which then has to be engraved or otherwise lettered, if you want nice labelled DIY.
|
Post by sadface on Nov 29, 2019 19:51:17 GMT 12
A few points: Firstly they can invoice in AUD which saves a fair amount in currency conversion compared to USD. 5x PCB up to 100mmx100mm is A$5.9. 10x PCB for A$7.8. 10x PCB with 2oz copper goes up to A$31 Funnily enough, I ordered most of these in black. Non-green generally doesn't cost extra, it just takes an extra couple of days. Green boards take 1 day. White, blue, red, white and black take 2-3 days. Non-green with 2oz copper goes up in price again as apparently its an unusual combination and therefore constitutes something like a special run. There is also an option for ENIG finish but that adds on the price even more. I ordered the LM338 boards in green colour with 2oz copper for current handling. The traces should be more than fat enough to handle up to 5A continuous. However as I understand, the traces assist with heatsinking for the regulators so I wanted the extra copper for dissipation. I intend to put these LM338 boards into use in a dual mono power amp using Rod Elliot's P3A to drive 4 ohm speakers to something like 100W. I also ordered the stacking rectifier boards in green simply so they match. Everything else I ordered in black as I think it will look nice. I will probably try using the ENIG finish when I get around to remaking my big tube preamp. The existing board is massive and is preventing me from doing everything I want to do in the chassis I have chosen for it. That's an A4 page for reference.
My evil plan is to clone(steal) the layout elements from the existing monolythic board but split it all up into 4 boards: 2x mono signal stage and 2x mono PSU. By splitting up the bits I should be able to fit a few extra bits into the chassis that I would like such as a relay source selector board.
I don't feel bad about stealing the layout because the supplier is a bit of a so and so. He takes any questions or issues with his kits as an attack on his person and his product and tends to respond with attacks rather than answers or solutions. Makes one not want to deal with him anymore.
The boards were ordered at 10pm on Tuesday, they are now completed and picked up by DHL. Damn these guys are good. That was including the extra time for black finish and 2oz copper.
|
Post by colinf on Nov 30, 2019 21:06:27 GMT 12
Looking good. I’m a little confused, are you using the NE5532 boards (also, you mean LME49720?) with the volume control as well as the valve board after it?
AMR-iFi R&D
|
Post by sadface on Dec 1, 2019 14:24:40 GMT 12
Hi Colin, The solid state preamp boards are for another project. A stand alone simple solid state preamp. I've designed it to use a NE5532 or an LME49720 per channel. Gain is 2.1x if I recall correctly.
The boards are in New Zealand. Now lets see how long customs and NZ couriers want to fluff about with them.
|
Post by sadface on Dec 2, 2019 19:57:23 GMT 12
Well,
Thanks to JLCpcb and DHL. The boards arrived this afternoon. 4 working days since I ordered last Tuesday night. The black came out lovely. Green is green like normal.
The 6N3 psu board
testing the mount holes for stacking the 6N3 psu on top of the regulated psu board for the LM1875
6N3 preamp board
The dual rail LM338 with its stacking rectifier board
Checking the mount holes for stacking
LM1875 big cap board
The Backup mono preamp board
Super low noise PSU boards for the Backup preamp Stacking low current speaker protection board for the backup preamp
Checking the mount holes for stacking.
Time to order some remaining parts and start building and testing.
|
Post by sadface on Dec 8, 2019 10:04:09 GMT 12
Well gents, One small hiccup so far. It turns out I used a bad pcb footprint for the tube sockets on the 6N3 preamp board. It is simply smaller than it should be for standard sized pcb mount noval sockets. I've managed to bend the pins on the sockets fit but I a bit disappointing at the necessity. C'est la vie.
I am wondering if I need to heatshrink the pins to prevent arcing now that they are closer than usual. However this might make testing voltages a bit more difficult.
|
Post by sadface on Dec 8, 2019 16:16:11 GMT 12
I had time for a bit of work today.
I got most of the resistors soldered on. I am waiting on the 510 ohm resistors for the cathodes on V2. Mouser will hopefully get them too me this week.
Vishay/Dale RN60 series resistors all round. I've generally ordered from RS components and Element 14. However I've discovered that mouser stocks a much wider range of values in the RN/CMF series, so it looks like I'll be using Mouser from now on for resistors. I find Element 14 is consistently cheaper for capacitors and silicon but doesn't stock a wide range of resistor values in appropriate quality.
Everything but the tube sockets will be mounted on the underside. The pcb will be mounted on spacers to the top of the chassis with the tubes poking out the top for venting. All of the power supply bits will mounted as far away as possible on the bottom of the chassis.
Lovely shiny joints is always gratifying. A while back I discovered the trick of using a Stanley knife blade to scrape the oxide layer off of the leads before soldering which made a huge difference to my joint quality.
|
Post by Owen Y on Dec 8, 2019 16:57:37 GMT 12
Noval socket - I don't see a possibility of 'arcing' across the socket pins, there appears to be plenty of fresh air between. Solder - are you using old type tin-lead rosin core?
|
Post by sadface on Dec 8, 2019 18:02:05 GMT 12
Hi Owen,
Yes, 60% tin, 40% lead rosin core solder. I've had bad experiences with non-lead solder. I was reading just last week about tin whiskers and the epidemic of early equipment failures due to shorting that is likely over the next decade or so since the EU regulations came in.
If I recall correctly the so called red ring of death that plagued the Xbox 360 was due to the transition from lead solder also.
|
Post by colinf on Dec 8, 2019 21:35:39 GMT 12
Yes tin whiskers might be a problem soon. I’ve been doing circuit boards that have copper pours have more clearance between the connections on it. So that the tin whiskers have to grow longer before causing a short circuit. Also been mounting circuit boards so that you can get to both sides easily so you can brush them every now and then, to cause stress to the thin tin whiskers so that they hopefully break off. Or just use tin-lead solder! Of course commercial products can’t use that here anymore.
AMR-iFi R&D
|
Post by sadface on Dec 23, 2019 13:11:44 GMT 12
G'day Gents,
Finally got all the last bits for the preamp board.
Here is the completed product.
Time to start working on the power supply.
|
Post by Owen Y on Dec 23, 2019 13:55:04 GMT 12
I used some of the JFX met. polyprop caps in a McIntosh MR71 tuner update last year. Like most wound film-caps, they exhibit polarity & ideally should be orientated appropriately in circuit for best performance - ie. 'outer foil' end connected to Gnd or to lowest impedance - for lowest induced noise/hum pickup. Importantly also, I've found noticeably more dynamic sound oriented this way. Assuming that these are coupling caps, connect 'outer foil to the 'driving' (plate) end. Unfortunately, you need probably a 'scope or millivolt meter to test ID the outer foil leads.
|
Post by sadface on Dec 23, 2019 14:38:45 GMT 12
Hi Owen,
Sadly, I am aware of this issue but have to trust blind luck this time around. I have been considering the need for a scope recently for this exact purpose: establishing the outside foil end of caps. It's a shame manufacturers don't mark the outside foil end any more.
|
Post by sadface on Jan 2, 2020 16:30:11 GMT 12
G'day Gents,
Happy New Year!
I've had a bit more arts and crafts time recently. The LM1875 boards are completed and ready for testing. A test power supply using one of the dual rail LM338 set to +/-24V is progressing along. I'm currently working on a modular chassis to mount all the various boards and transformers for testing. This has come up the priority list as its holding up the Backup Preamp.
Here are the completed LM1875 boards: I love spade connections!
Notice the space in between the 2 tall 470uF caps to allow access to the IC with a screwdriver. This feature and the fuse rails were missing in the cheap ebay boards I started with.
And here is the PSU board progressing along:
|