Monday, 31 October 2016

Since we're building power supplies...

Following the success of the main power supply for the amp, I decided to improve the USB power supply I'd previously built on a piece of stripboard. The idea of the USB supply is to power the Audio Chromecast which will be feeding into this amplifier. Thus saving the need for a wall-wart.

What's needed is an isolated regulated smooth 5V DC power supply – so a simple circuit powered by the heater AC, containing a rectifier, some capacitors, and a voltage regulator, is all that's needed.

So utilising my newly-acquired PCB design skills, I decided to make my own PCB for this and build the circuit improving on the previous effort. This is the result (the SD card is just a scale prop, since the photos of the previous supply didn't have one I thought it might be a good idea this time)

A bit smaller than the last one.

Tested and working quite happily, it powers the Chromecast and the DC is very smooth.

This is the PCB for it:

Actual size 45 x 35 mm

And changing the subject completely, I thought it might be a good idea to publish a new circuit diagram, since the previous one on here is well out of date by now.
You'll need to click this to see full-size

The circuit diagram however does not show the delay circuit I built with the 555 timer and the relays, so the power supply here is "conceptual" rather than as-built. The relays switch the AC secondaries going into the rectifiers.

Now, the desk-bound work continues - designing the layout of the top, front and back panels, for the CNC machine to cut out.

More later...

Wednesday, 26 October 2016

Testing the power supply

So I put the power supply in a test chassis and wired it up, this video shows the first switch-on.

Did it work? Was there smoke or fireworks? Watch and see...

Power Supply board, Part 2

After the previous less-than-completely-successful circuit board fabrication effort, we now have a board that is 100% correct. I am using the "press 'n peel" method of printing the layout onto a plastic sheet then using a hot iron to transfer the design onto the copper side of the PCB, before then putting the PCB into the Ammonium Persulphate etchant to dissolve away all the copper not covered by the resist (=transferred printer toner)

The transfer process was time-consuming and not particularly effective, it required multiple passes with the iron and then careful peeling off to see how much of the resist was transferred before then carefully re-ironing.... rinse and repeat several times until it seemed no more resist would be transferred. Then it was a case of filling in the missing parts with a sharpie, fortunately not much was needed.

From there it was into the chemical bath for 10 mins being constantly agitated until all the remaining copper was dissolved off.

Then... drilling, cleaning, then stuffing and soldering.

The stuffing and soldering took around 2 hours, this included testing the delay circuit to make sure the timer circuit (my design) was working. Happily it was.

The timer circuit takes the 6.3v AC heater supply, feeds it through a bridge and then into a 5V regulator and then to a 555 Timer IC with suitable component values to give around a 30sec trigger delay before closing the relays and applying the high-voltage AC to the rectifier diodes. This is to give the valves a chance to heat up and be ready to draw current before any HT is applied.

This is the completed supply:

On this board we have:
  • Timer delay circuit including relays
  • Rectifier (Diodes in series because of peak inverse voltage exceedance) plus balancing capacitors
  • Main capacitors (4x, in series-parallel configuration, with balancing resistors)
  • Separate supply for input valves (x2, 300V)
  • Separate supply for Driver valves (x2, 360V)
  • Separate supply for Negative bias voltage for output valves (X4, -80V)

Next step: testing

Monday, 24 October 2016

Power Supply board, part 1

Following on from my previous post about the power supply board, I made a number of improvements to the layout since the initial design. Also I completed it by adding the bias supply components.

Confident I had something I was ready to produce, I went through the process and learning curve of making a PCB. This resulted in some difficulty transferring the design successfully from the plastic film it was printed on, to the board ready to etch.

However a few trials and errors and finally I had this ready to start stuffing, after carefully etching and drilling:

Ain't it beautiful?

The first usable PCB I ever made, or at least, almost usable... 

On this board is a galactically stupid n00b mistake which renders it completely useless: I'd neglected to mirror-image the pinout of the 555 IC I'm using as a start-delay. So in order to get this working I'd need to mount the IC to the underside (ie. solder-side) of the board.

Not happy with that at all. 

So I've corrected the design and it's back to the electronics shop, can't go tomorrow as it's a public holiday, so now need to curse my naivety until Tuesday when I can get another board and transfer the design onto it one more time...

THEN I will have a power supply board that works...

However this experience has taught me a few useful things. Firstly, the heat transfer method of getting an image from a laser printer onto a board is a bit hit-and-miss, but I found polishing the board with Brasso first then roughing it up with steel wool helped.

Secondly, despite this board having 110 holes (or thereabouts) the drilling process did not take long, the drill stand makes it easy.

Thirdly, sharpies are magic for filling little holes in the resist prior to etching the board.

Tuesday, 18 October 2016

The volume control that comes with unintended consequences

So I finally got my hands on a decent-ish signal generator. A few online reviews of this device haven't been particularly kind to it, but the shortcomings discovered are well outside the range that I'll be using it at anyway. For my purposes it's 5-stars excellent. 

This is the device, an AliExpress-special, it's called a "MHS-5200A":

It hits all the high notes the previous two signal generators I was using couldn't... so today I was able to test my response out to 100kHz.

Which revealed something extremely interesting... my volume control, which is simply an attenuator on the input, is having an effect on frequency response.

Not entirely surprising I suppose, given that the combined effect of the volume control and the capacitance of the valve would form an RC network.

So to recap, this is the relevant section of the circuit:

With a sig generator that will easily give 100kHz (and more) and at a nice 200mV RMS (which the previous couldn't) it means I can test the frequency response with the attenuator at maximum (equivalent to maximum volume on a finished amplifier) as well as with the signal being attenuated.

The test was taken from the output of the driver stage - in other words, the signal went in the initial gain stage, through the concertina, and then through the driver, before going into a X100 probe and then to the oscilloscope.

The result was interesting... and confirmed my suspicions. Volume controls come with hidden extra features. Check freq response graph below... I tested at 5kHz intervals from 1kHz up to 100kHz, so this graph doesn't contain any extrapolations.

So the green trace is the freq response out to 100kHz with the volume control set to maximum and a 200mV RMS input. I had a probe on the input as well and my spreadsheet normalized for slight variation in signal voltage on the input (verified as being the same whether connected to the amp or running open-circuit)

The red trace shows the response with the volume control at -10dB... so the signal is going through some carbon track.

Probably not enough to make an audible difference, but interesting and informative.

I'm also quite rather pleased that my naive circuit built using valves that are everyday quality rather than audiophile spec, managed to turn in such a good response with zero attenuation dialled in.

Saturday, 15 October 2016

Additional measurements, power supply design

In the last couple of weeks I've spent a lot of time agonizing over how on Earth I am going to make a chassis for this project. I have zero sheet metal skills and very little equipment. I vacillate between buying a ready made chassis from AliExpress and somehow cutting all the holes I need in it, to ambitiously planning a magnificent construction with corner braces and a polished aluminium top plate.

Fair to say this is not resolved yet and will be the main factor holding me back now.

However on a happier note, I've decided that the power supply circuitry will be going onto a PCB. This is necessitated designing a layout, which I have been doing completely the old way - manually. Measuring component footprints and putting the pads on manually.

Those who are more seasoned at such things will observe my technique and doubtless recoil in horror in much the same way you might when you see Grandma transcribing an email because she doesn't know about copy-and-paste. However with nobody to get me started or show me how to do this, I've resorted to the electronic version of essentially manual sketching.

So on this power supply board - which is 150mm X 150mm (and once again seasoned professionals at this type of work could probably fit all this onto a board quarter the size!) will be the rectifiers and main capacitors (4 of) plus the voltage droppers and capacitors for the 4 input valves (one each) plus the output valves bias supply. As well, there is a delay circuit consisting of a 555 timer IC and the necessary discrete electronics, to switch on the HT supply (on the AC side) after approx. 30sec warm-up time for the valves.

This circuit will be fed with a rectified and regulated 5V DC sourced from the heater supply. An LM2940 5V regulator will be used, solid state is allowed in this amplifier, just not in the signal path!

The PCB layout isn't quite finished yet as I still need to add the bias supply. Those of a professional disposition may wish to avert the eyes at this point. This is what the layout looks like so far:

As you may deduce, this is the first PCB I have ever designed

If anyone's got a genuine "Warning - if you build that, then XYZ will happen!" to contribute, please let me know. But if you just want to criticize how crap my design is, kindly please don't. As long as it works, I will be happy, and I am not mass-producing these things. 

I will post photos of this board from both sides once I've stuffed it.

Yes the main caps are in series, this is intended, yes I know that series connecting electrolytic caps is weird. There will be balancing resistors across them... it's just that the caps are 450V rated and my B+ is 560V.

Moving on, I have also sourced a signal generator that goes to 65 kHz, the previous one I was using had a max frequency of 20kHz. So I have made some more measurements of the test rig on the bench - this is the input, splitter and driver stage. The -3dB point measures at around 50kHz. With the cheerful assistance of my daughter as a lab assistant entering numbers into the spreadsheet that I called out as I made measurements, this is the response I am getting:

Which I am also reasonably happy with.

In other news
 - the KT88 output valves have arrived
 - I am still waiting for the remote volume/input switcher from AliExpress
 - and I still haven't figured out what to do about the chassis!