Monday, 29 August 2016

Milestone 1: Completed Circuit

A milestone: The circuit diagram is complete, at least the basics. This doesn't include the funky bits like the input switching module - this diagram is just for the amplifier proper.

I've completed the power supply, negative feedback, and bias adjustments for the KT88s.

This is back on the forum for the wise folks there to evaluate and provide their robust feedback.

Next steps: shopping for parts (the Mains transformer is already to hand, output transformers ordered, resistors will mostly come from stock... everything else will need to be ordered.

STILL haven't got any further with what I am going to use for a Chassis. Two conflicting forces here: The desire for acceptable aesthetics and appearance, and the strength to hold around 22kg of iron (Each OPT is 5.5 kg, the mains transformer 11kg)

Click for enlarged view.

Friday, 26 August 2016

Slow progress

A couple of weeks of urgent family matters that needed attending to have put the brakes on the amp design, however things haven't slowed to a complete freeze.

Around one week ago, my custom-made power transformer arrived, and it is a beast. Just over 11kg and quite imposing on the bench

The digital caliper is just there for a size reference.

On this transformer we have a 230v primary
secondaries are:

  • 3.15 - 0 - 3.15v @ 8A (heaters)
  • 60v @ 200mA (for KT88 bias supply)
  • 400 - 0 - 400v @ 500mA (main HT supply)

Should be sufficient for the purpose. The challenge is going to be how to make it look nice in the chassis - it's a bit... naked.

Moving on, we are also near completion on the circuit diagram for the amp. A lot has been added and changed since the diagram was previously posted here:

(Click to see larger size)

  • Changed Circuit diagramming software to Tinycad (and made up my own parts library for the transformers)
  • Added KT88 final stage
  • Re-arranged 12AU7 driver stage layout in the diagram
  • Changed all resistors to preferred values, recalculated bias and voltages as required
  • Added theoretical voltage reference points through the circuit
  • Calculated power dissipation across all resistors
  • Calculated DC voltage across all capacitors
  • Added voltage dividers to power the gain, splitter and driver stages (modelled in PSUD2)
  • Added power supply (not yet completed)

Still needing to be done:

  • Add adjustable bias supply circuitry for the output valves
  • Add the negative feedback section

And as an additional safety feature to protect the KT88s, I have decided to have the 500V B+ switched through relays which are activated by the bias supply (one relay per valve). This way, if something catastrophic happens like a pot wiper lifting or a resistor burning out in the bias supply, the relays will open and the B+ will shut down.

Another subject I have been musing is the shutdown behaviour. There's going to be 2000µF of capacitance on the main HT which means it may take a few seconds to shut down when the amp is powered off. Potentially there may be a situation where the some of the B+ HT is still at the anodes of the output valves while the bias supply has shut down.

Given that there's no load on the bias supply this is unlikely, however it is a point I want to measure carefully in the construction.

Once I've completed the design, I'll post the circuit up on the forum once again and see what I get in the way of constructive criticism.

Sunday, 14 August 2016

Dummy Load

All of the work so far in the design of my amplifier has been away from the workbench... all the reading and increasing knowledge, and all the circuit design work, take place at the computer or elsewhere.

Currently I have no parts at all for the amplifier yet. The mains transformer is being made, and when that arrives, it'll be the first actual component of the amplifier to arrive through the door. So I am a long way from actually starting to build.

Apart from designing the circuit, the next challenge I face before I can start building is that I currently have no idea how I am going to find a chassis. When I do, I need to then work out the physical layout of the components on the chassis, draw that out (so adding a new skill... CAD drawing, which I have never done)... then send it off to the sheet metal workshop to get all the holes cut.

Then I've *somehow* got to make that chassis pretty enough to withstand being on display and (dare I flatter myself) even admired, in the pleasant environment of the living room. No idea how I'm going to do that yet.

So in summary, lots of not much going on at the moment. A test of the patience!

In order to satisfy my action bias in an area that has little scope for disaster, I decided to make up a simple dummy load, since this will be needed when testing the amplifier. Essentially this is like a large box of hammers and about as dumb... what could be simpler than a bunch of wirewound resistors in parallel, that has to dissipate a bit of heat.

This will be an inductive load, so like a real-world loudspeaker, in other words...– if an amplifier can't be stable powering a slightly inductive dummy load then it won't be stable powering a loudspeaker

This one's made up of 10 x 47ohm 10w resistors in parallel. It measures at 4.5ohm. Due to its common-ground construction, it can be used as a 2 channel 4 ohm load or a single-channel 8 ohm load.

To test it I connected a 12 volt SLA battery and observed the (lack of) fireworks. The resistors got warm as expected but no drama.

12 volts 4.5ohms equals 32 watts of heat being disspiated.

then just to give it a bit more to think about I connected another 12V battery in series with the first. Now we have 24v going into 4.5ohm – double the voltage and hence double the current – so 4 times the power. Giving a total of 128 watts. Now the resistors got too hot to touch and a few small wisps of smoke appeared as bits of stray dust burned off, but the smoke cleared and the load appeared to be OK, after three minutes I removed the power and re-assembled the case as the resistors cooled down.

128 watts is more than I would ever anticipate putting into the load from my amplifier, so I feel this load will probably serve the purpose, unless the inductance becomes a problem, in which case a stronger single resistor will be called for.

The dummy load, minus lid:

Not gonna cut it if I start designing kilowatt amplifiers, but I don't think that will happen any time soon. The lid of the case is vented and I could add a cooling fan if needed, or else re-design to mount all the resistors on a heatsink. However, this should be adequate for the 50w or so it's going to be called on to disspiate.

Now I don't have anything much else I can build until the amp itself. :(

Thursday, 11 August 2016

Reading list

The reading list is growing. As mentioned in the previous post, there's the manuals for the oscilloscope, and today on the recommendation of the community in the Valves forum on DIYAudio, I've added these two:

Valve Amplifiers, Fourth Edition, Morgan Jones

Building Valve Amplifiers, Morgan Jones

Duly downloaded onto the Samsung tablet.

Meanwhile to keep the construction going, I'm currently building the dummy load I'll need when testing the amplifier.

Wednesday, 10 August 2016


So my USB oscilloscope that I ordered from AliExpress arrived today. I set it up and did some basic measurements using the signal generating app on my Android tablet.

1kHz sine wave looks OK

Square wave however is a different matter:

Also, the PDF manual contains some fantastic English translations that almost make sense:

"Note: The specific speed recorder with computer processing speed, and if use high sampling rate, the situation may break."


"Note: The oscilloscope factory calibration, if you are not satisfied with the measurements, canmanual calibration, the specific reference oscilloscope instructions."

This scope will be a vital piece of test equipment to let me see what's going on inside the amplifier during building.

Next task: reading through the manuals.

Tuesday, 9 August 2016

Peer-review burns

In the Academic world, when you submit your work for peer-review, you can expect some criticism. If your work contains errors or draws a conclusion based on a misconception, you can expect that critique to be robust, to use a euphemism.

In the case of my amplifier design, I decided to use the forums in DIYAudio as my council of peers. Since this is the first amplifier I've ever designed, as expected my initial design contained a few omissions, and I'd fallen for a few Gotchas. The community were initially dismissive in their critique, but pressing the goodwill I continued being friendly and thankful for the information, and then some useful suggestions began to turn up.

So - as a result of that helpful advice, below is the amended version of the initial stages of the amp.

Note that the topology is unchanged: 12AX7 initial gain stage using Grounded Cathode, 12AX7 Concertina phase splitter to follow, followed by a 12AU7 gain/driver stage. However there have been some changes:

Click the image for full-size view

First: The volume control has been reconfigured. I had read that using the volume control as a shunt to ground resulted in less noise as the pot wasn't in the signal path. This suggestion was robustly challenged, so I've adjusted the configuration accordingly.

Second: The Cathode bypass capacitors. I was using the values TubeCad had given me but it was suggested these were too low. Researching this further it seems that the exact capacitance needed is not easily defined but too little is much worse than too much. So I increased them

Third, the most grievous error I'd made... the second half of the 12AX7 – the concertina phase splitter – I'd completely omitted to bias the grid. Fatal error! So now there's a voltage divider network of resistors to get the grid about 2V below the cathode. These resistor values may need fine-tuning by doing measurements in the circuit.

Fourth - the 12AU7 drive stage - adjusted anode resistor value and tied the cathodes and gave them a common cathode resistor and capacitor

I'm going to leave this alone now but at the time I begin building the circuit, I expect that some modifications by trial-and-error will be necessary, using actual measurements and traces on the oscilloscope.

Next I shall move on to the finals and then the power supply.

Monday, 8 August 2016

First component built

While the circuit for the amp is in the design phase I've decided to prepare one of the simpler stand-alone components - this is strictly a "nice-to-have" but I've decided that the amp will have a 5V USB socket on the back for running a USB-powered source accessory.

Currently I'm running a Sony integrated, which the valve amp will eventually replace. Connected to one of the inputs on this is a Chromecast Audio, which I'll run into the valve amp when it's built.

So with this in mind, I decided that a simple voltage regulator circuit running off the heater supply would do the job. However most linear regulators have a 2v dropout voltage, meaning that the input voltage must be at least 2v higher than the output.

However, the LM2940CT-5 regulator has a 0.5v dropout voltage, which is handy because when putting a bridge rectifier across the 6.3V I'll be seeing just a shade over 6V DC.

So, a bridge, a handful of capacitors, a voltage regulator and a heatsink, and a USB A socket.. assembled in the correct order, should give us a nice clean stable 5V of USB power.

This is the finished article – it doesn't look beautiful, it's just made on a piece of stripboard. But it's been tested and works 100% fine with the Chromecast.

(Finished as in electrically finished... I'll probably do something nice about the rough edge on the stripboard, and maybe I'll get a flush-mount USB socket rather than using the socket right on the board. OR maybe I won't even bother if space is too tight, and just keep the Chromecast on its own external supply. Like I said, this is a nice-to-have rather than an integral component!)

Wednesday, 3 August 2016

We have a preamp stage

Progress in the designing phase - we have a preamp stage.

The topology is simple: Grounded Cathode initial gain stage using half of a 12AX7, giving approx. 50x gain, AC coupled to a split-load (or Cathodyne) phase splitter using the other half of the 12AX7. This section has unity gain.

From there, each phase of the signal is AC coupled to half of a 12AU7 to give the necessary voltage gain to drive the KT88s I plan to use on the output.

The 12AU7 driver stage should give around 10x gain.

This is the preamp design:

Click the image for a readable size

I could have done this the easy way and just stolen a circuit off the internet. But my objective was to learn something, not just assemble a jigsaw puzzle. So I have spent many hours in the last week reading about bias points, gain, grid resistors, different phase splitter topologies, etc...then I spent a long time modeling in TubeCad and then double-checking with the calculator. This means I know what's going on in this circuit (well, that's the theory anyway)

So we should have enough gain here to power the KT88s to full power from a 100mV input signal at the grid of the first valve.

It will be very interesting to build this and see how far my measurements differ from the theoreticals predicted by TubeCad and my calculations.

These are the bias points I've chosen

Bias Point for 12AX7 Initial Gain stage

 Bias Point for 12AU7 Driver stage

Next up - a peer-review of my preamp stage, then on to designing the output. Stay Tuned...

Tuesday, 2 August 2016

Shopping begins

So this being the first amplifier I've designed and built, there are a few tools I'm going to need.

Firstly, a decent quality voltmeter - better than the cheapie I'm currently using, and most importantly something rated to handle the voltages I'll be measuring – up to 500V DC and 400V AC.

Secondly - a signal generator. Very necessary for putting test signals into the amp to determine if it's doing what it should. Normally these things can cost tens to several hundred dollars, depending on how full-featured and robustly built you want them. In my case since I am only interested in audio frequencies, an app on the Android tablet will serve the purpose

Keuwlsoft Function Generator - Free and without advertising.

Next: An oscilloscope. After looking at the options available I've decided on a dual-channel "soft" oscilloscope - one that uses a laptop for the control and display, 

Instrustar Oscilloscope hardware - the rest of it's in software.

Finally, I've also got a power transformer on the way. After evaluating all of the commercial off-the-shelf products and becoming increasingly dismayed by the cost, I decided to see if a local manufacturer could make me one based on my spec.

Luckily they could, and for about half the price of buying one from an overseas manufacturer. The power transformer is the single largest and heaviest item in the whole project, I expect it to weigh in at around 10-12 kg.

This was the spec I passed to the manufacturer:

Handily drawn on my Surface Pro. I love that computer!

Luckily this was enough of a spec for them and the transformer will be wound and sent to me, expected arrival 2 to 3 weeks.

Next steps: Finish designing the preamp stage. I've decided on the topology and just need to draw up the circuit, which involves cursing at the rather primitive diagramming software. I'll post that up next.