Post by sadface on Jan 8, 2020 20:31:56 GMT 12
G'day folks,
I finally got the amp boards mounted on a heatsink. There was a few false starts but the third time was lucky.
I learned that even though my tap states m3x0.5. It is in fact a tap for thread repair on m3x0.5 and therefore requires the use of the little spring insert things.
It took me 2 attempts to figure this one out. My first attempt was done with a cordless drill and therefore wandered all over the place. 2nd attempt used the drill press at work. 3rd attempt used an actual m3x0.5 tap.
After the first 2 failures, I didn't have a heck of a lot of room left on the heat sink to fit both boards around the faulty tap holes: The last set of holes were offset 5mm from the 2nd set and therefore there is a bit of material missing on the heat sink under the chips.
I have my concerns about issues with thermal coupling. However I'm going to run with what I've got for now. If i encounter heat issues later I will know where to look first.
Here's my test fit minus the insulation tabs.
One can see my first attempt that wandered horribly with the cordless drill at the top (if you look carefully at the top left hole one can see how far the centre of the hole wandered from the horizontal score mark) , the 2nd set of holes can be seen on the outside of the chips.
The heat sink polished up fairly nice. After drilling and tapping etc, I cleaned it with a combination of paper towels, makeup wipes, iso-propyl alcohol and pcb cleaner. It amazed me how many times I needed to wipe it to get rid of the cutting oil and swarf.
I sanded it from 80-220 grit, then went to some 1200 grit I had lying around, then a synthetic steel wool finish pad of unknown grade.
Now I just need to get around the sheet metal work on the testing jig. Many holes have been drilled, now its the hard part where I have to cut out some sockets for an IEC socket, fuse holder, power switch and I/O panel. I also need to remember to drill some holes for feet that I never seem to remember.
The idea here is a test jig to accommodate the various board footprints and transformers I'm playing with right now.
I see people's test setups and they often look more than a little hazardous. I want a test jig that is electrically safe and doesn't look like a birds nest as I generally like to avoid electrocution.
The marked out sections are for material removal. Circled holes require further drilling out to proper size. The 4 straight lines are simply to highlight my score marks for where I intended to bend the sheet into an open topped U shape with all the usual stuff at the front and back.
Bending might be fun: I've previously bent ali sheet over the corner of my work bench using a bit of wood on top and some quick clamps. I'm not sure if this is going to work with the 16gauge ali sheet in front of me but I'll deal with that once I get there.
|
Post by Owen Y on Jan 9, 2020 9:04:09 GMT 12
Bending metal - sometimes I use a B&D Workmate & clamp the sheet between, then hammer it into the 90deg bend required. (Soft hammer and/or block of wood to avoid surface damage.) If you want a sharper fold, you could use a square-edged pc of wood or two, with sheetmetal between, clamped to a bench edge (as you say) & do similar. Or, if you have one, a bench vice & soft hammer.
|
Post by sadface on Jan 10, 2020 21:53:16 GMT 12
G'day chaps,
I had a bit of time tonight.
The dual rail LM338 psu has been completed. Set to +/- 24vdc.
The rectifier boards have a small design flaw: I made the hole diameter for the MUR820G diodes a bit too large so they don't sit on the 'shoulders' of the pins. I had a bit of fun trying to keep the chips sufficiently proud of the board. I wasn't entirely successful.
At least with this build their not going to pass more than 2.5A so I am hoping heat won't be an issue.
I do have some small heat sinks I can apply if necessary. I am too lazy to do the maths, I'm just going to assume for now that since a diode like a 1N4004 can pass 4A without a heat sink the MUR820 should be able too as well.
I will buy a infrared temp gauge to keep on eye on things.
The blue caps are 0.1uF class X2 for noise suppression on the mains side. The ceramic caps are 220pF class C0G for noise suppression on the diodes.
I am also slightly concerned about how close the reservoir caps ended up to the heat sinks. On future use of these boards I might be able to get an extra mm or 2 of clearance between the caps and the heat sinks by mounting them as far back in the holes as possible.
Later on I would like to make a 50wpc discrete amp (I have some P3A boards from ESP for this) powered by one of these regulated supplies each channel. I am a little dubious about these heat sinks being able to handle 5A without overheating.
It might require off board heat sinks or simply a different design such as the mosfet psu from earlier in the thread which spreads the heat over double the devices. Ultimately, this current build is a test run for the bigger discrete version.
By my maths; I currently run 4ohm Dynaudios. Meaning 100w instead of 50w.
100W/4 Ohms = 28.3 V peak / 7.07 Amperes peak. I plan to use a 30vac transformer @ 5A per channel.
As far as I understand I want about 8-10v drop over the regs to prevent sag and provide headroom for peaks. Meaning at the upper end of usage the regs could be dissipating 50w - 2.6 C/W heat sinks just don't sound sufficient. I'll cross the bridge once I get to it. I'd be running them at +/- 30vdc which will probably limit output slightly. There is also the question of whether I will ever use 100W, I wouldn't be surprised if under an actual use case I never use more than 20w and the LM338 boards will be fine. Time will tell.
|
Post by Owen Y on Jan 11, 2020 8:27:38 GMT 12
The rectifier boards have a small design flaw: I made the hole diameter for the MUR820G diodes a bit too large so they don't sit on the 'shoulders' of the pins. I had a bit of fun trying to keep the chips sufficiently proud of the board. I wasn't entirely successful. Hey sadface - you may have done this, but maybe you can slightly bend the centre pin of the MURs, to get a tighter 'interference' fit in the bigger holes. Thus allowing you to lift them off the PCB. Maybe it doesn't matter too much anyway, as the metal tab is up a bit...?
|
Post by sadface on Jan 11, 2020 9:10:44 GMT 12
Hi Owen. Thank for the suggestion. An obvious solution that didn't occur to me whatsoever. These are a 2 pin device, but the advice is good nonetheless. If I was to bend the 2 pins outward or inward slightly, I could get them to rest properly on the shoulders.
I'll remember that when I build some for +/- 30vdc.
|
Post by Owen Y on Jan 11, 2020 10:10:08 GMT 12
Ah yes of course, diodes are 2-pin By my maths; I currently run 4ohm Dynaudios. Meaning 100w instead of 50w.
100W/4 Ohms = 28.3 V peak / 7.07 Amperes peak. I plan to use a 30vac transformer @ 5A per channel.
As far as I understand I want about 8-10v drop over the regs to prevent sag and provide headroom for peaks. Meaning at the upper end of usage the regs could be dissipating 50w - 2.6 C/W heat sinks just don't sound sufficient. I'll cross the bridge once I get to it. Some thoughts (bearing in mind I'm a novice in this SS area)... - Depending on the speaker imped curve, 4 ohm may be a minimum dip somewhere other than at LF end, which is where the bulk of power demand will be. - Allowing for power headroom say 15-20dB at least, the amp should never be asked to deliver more than10-20%(max) of available power. In which case a 2.6C/W sink may work OK if you can keep internal ambient temp down - ie say heatsink 25deg above ambient / say 10W dissip. They do look super close to the caps, but good ventilation may be your best strategy.
|
Post by sadface on Jan 11, 2020 12:39:17 GMT 12
Hi Owen,
I'm also a bit of a novice with solid state (less of a novice with tubes but still a novice nonetheless).
I've not done the maths on it. But I've been somewhat relying theory of only using 10-20% if available power in normal operation.
From what I can tell from the specs on my Dyns they are 3.8ohms at around 200hz. Well above the crossover point on my active crossover which sucks.
100w is still a heck of a lot though. My tube mono blocks are good for 45w per channel in the current configuration and that is more than enough for me.
|
Post by sadface on Jan 11, 2020 16:13:52 GMT 12
G'day Boys,
After some open homes this morning (not promising) I've had the afternoon to get the sheet metal work done.
I used a dremel with a 1.5" metal cutting wheel to make a slot and then used a jigsaw to cut out the rectangular sections. There was a tiny bit of over cut on the small power switch socket but meh.
Its been a very good day. I had an excuse to use 4 different power tools so far....
The last holes drilled out to size.
My high tech bender.
Crude but functionally sufficient for what its going to do. I just a testing jig afterall. I did make a funny mistake and bent it the wrong way. Now the layout is reversed. For some reason I always want my power supply on the left side and signal stages on the right side. This worked out backwards.
Not a big one but its probably going to tweak my OCD when I look at it I will need to drill an extra hole on the other side of the volume pot socket since the locating hole thing is now on the wrong side.
Now its onto wiring up the mains side. I/O and then I can test out the Backup Preamp.
|
Post by Owen Y on Jan 11, 2020 19:57:00 GMT 12
For me, the fun part is the soldering, but you invariably have all the time-consuming hardware parts to labour through first
|
Post by sadface on Jan 11, 2020 20:14:25 GMT 12
Hi Owen,
Soldering is my favourite bit also.
I do however thoroughly enjoy anything that involves an impact driver or a saw. Drills are ok.
Wiring is probably my least favourite bit.
|
Post by Owen Y on Jan 12, 2020 16:05:32 GMT 12
Wiring is one of my favourite bits - I like to try emulate those incredible Japanese DIY builders in MJ magazine Woodworking I much prefer over metalwork.
|
Post by sadface on Jan 12, 2020 20:41:05 GMT 12
Good evening guys,
I got the test jig setup. The old measure twice cut once bit me again: I didn't account for enough transformer bulk inside the chassis and thus had to move my power switch socket over 35mm for clearance. Good thing its only a testing jig and doesn't have to look nice.
I'm pretty sure I have a handle on safe wiring practices however if anybody spots anything untoward please let me know.
Everything is setup so that I can test the psu and relay protection boards for the Backup Preamp from my other thread. Unfortunately, I seem to have lost or thrown out the fuses I got from Jaycar last week.
No testing of the new preamp for me tonight I'll have to swing by Jaycar again tomorrow and pickup some more fuses so that I can get onto testing tomorrow night.
I was intending to use a 0.1A slow blow fuse. However I installed a 3AG fuse holder and Jaycar only offers a quick blow fuse in a 3AG package.
I'll try my luck with a 100mA quick blow however I imagine that I might blow them with the inrush current on the 500mA transformer so I'll buy some slightly larger values as a contingency.
Power draw shouldn't be more than about 80mA for both preamp boards and the protection board by my maths.
|
Post by Owen Y on Jan 13, 2020 8:30:48 GMT 12
I tend to 'breadboard' onto wood. It's quick & easy to screw into, etc. This was an early mess ...
|
Post by sadface on Mar 4, 2020 21:06:22 GMT 12
G'day Gentlemen,
An update on this project.
I've been beavering away making a clone of the Sigma 22 Power supply board. I would have liked to buy the genuine article but the shipping on them is quite prohibitive and I ultimately intend to use a few of these.
I have an ebay version of it, however I've decided to make my own clone off some good quality photos of the original boards.
I've made a slight modifcation to the dimensions to accomodate 35mm filter caps in place of the 30mm that is available on both the bay boards and the original.
Here's the schematic again.
And here is the bottom side ready to go.
The top side might take me a wee while longer and then some fettling of the silk screen and it should be ready to go.
|
Post by sadface on Mar 7, 2020 12:30:58 GMT 12
G'day,
This took much longer than it should have. EasyEDA has been running very slow for me due to the way I was trying to use it.
Here's the completed PSU board. A clone of the AMB labs sigma 22 with a few modifications to suit how I like to build things: Spade connections and larger footprint filter caps mainly.
and here's the original that I cloned off of this photo.
Now I just need to make some pcbs to house the output relays and I can finally test out the LM1875 boards.
|
Post by colinf on Mar 8, 2020 7:33:50 GMT 12
Nice one. What was causing Easy EDA to be slow? I tried it a while back and found it a bit clumsy to use, although it is free which is good! I used DesignSpark and then upgraded to the Pro version which is more precise with better routing options. Can you use Library Loader with Easy EDA?
AMR-iFi R&D
|
Post by sadface on Mar 8, 2020 11:54:31 GMT 12
I think, I have in essence created too many objects to make my ground planes.
I was handling fills by coping and pasting sections of wire. I was also copying and pasting small sections of wire to handle fine routing around objects and wires instead of using continuous wires and the fill tool.
The more I filled in the ground plane, the slower it got.
On the top side I retrained myself to use the wire tool more effectively and also to use the fill tool. Even with the software running slow it was much faster than the way I did the bottom side.
A week or 2 back I got frustrated enough to download KiCad. I spent an afternoon trying to copy over my designs from easyEDA to Kicad. One can convert the pcb layout over, however not the schematic files or the nets so I gave up.
|
Post by colinf on Mar 8, 2020 20:13:13 GMT 12
Yes it can be very confusing, each program has its way of working that you have to learn. Not tried KiCad, a mate uses Pads and swears by it. DS Pro had a steep learning curve as well but after you work out it’s quirks (and that takes days of frustration!) it’s fast to use. Although I don’t use it much now as work needs Altium Designer, a full featured program, priced accordingly!
AMR-iFi R&D
|
Post by sadface on Jun 13, 2020 19:12:39 GMT 12
G'day Guys,
The boombox was raided for parts over the lockdown. Now She Who Must Be Obeyed wants the boombox back so I've started on the rebuild.
I think this must be something like a version 2.0.
A while back I removed the old linear PSU board and replaced it with a +/- 24Vdc regulated PSU using LM338 ICs. The dual LM338 psu works great. I've tested it at a decent volume in the warehouse at work and the heatsinks were at 29degrees.
Its finally time to try out my 'Big Cap' LM1875 boards.
Last time they were seen they sported Wima 3.3uF MKS2 polyester caps on the input.
I've since upgraded to some Panasonic 2.2uF polypropylene types. I can't say I've heard the MKS2 caps in this implementation. However experience suggests that polyester caps sound a bit harsh compared to polypropylene so I made the switch as they Panasonics were in the parts box.
The boards are now properly mounted to the heatsink with insulation pads and Arctic Silver thermal grease.
Later tonight I should have some more time to rule up the mount holes for the boards and heatsink.
After that I simply need to solder in some new RCA sockets and wire everything back up for testing.
Eventually I will get around to finalising the design of the enclosure and speakers into something less crude however for now plain old serviceable will have to do.
|
Post by sadface on Jun 14, 2020 20:54:25 GMT 12
May I present the completed but untested version 2.0.
Unfortunately I am lacking fuses to test it out tonight but its ready to go after a visit to jaycar tomorrow.
I think that I need to do another revision of my dual LM338 board with an extra set of outputs and some extra ground tabs.
|
Post by sadface on Jun 14, 2020 21:08:34 GMT 12
I will probably end up abandoning the monolythic boombox idea. I've been researching speaker designs for a while. I'm especially interesting in trying out some good full range drivers however I'm yet to find a design that is both compact enough to be integrated into a boombox and with sufficient bass response. With this in mind I'm starting to head in the direction of just making some decent sized bookshelves and a separate integrated amp. Somewhere between China and NZ is a bluetooth receiver board I ordered back in April which will make operation easier.
Possible designs I am looking seriously at are the new Mark Audio CHN-110
I'd prefer a front ported design however as they would likely end up against a wall since its for the bedroom
|
Post by sadface on Jun 15, 2020 22:17:52 GMT 12
It Verks! No smoke, no explosions, no hum. Just a nice smooth sound. I will let this play for a couple of weeks to figure out its sound. Then I will start integrating the tube preamp stage. Its ugly, but give me a year or two and it will turn into something worthy of the shelf space.
|
Post by Owen Y on Jun 16, 2020 10:18:43 GMT 12
The main thing, the first thing, is that it works well
|