Junque Box Boutique Amplifier.

After repairing Norman's Rickenbacker amplifier, he came to me and said "Andy, you can't keep repairing guitar amplifiers if you don't own a guitar"

Fair point.

So he very generously provided me with such.

Er ...

... yeah ...



So, after repairing the recent Marshall amp, I managed to wrangle the opening bar or so of "Matchstick Men" by Status Quo out of it, then Steve came along and collected the Marshall, much to the relief of the neighbours.






What's needed here is an amp to call my own. Shall I be going out and purchasing such? No. Shall I be looking round in the workshop to see what we've got kicking about that I can cobble together at minimal cost? Read on ....

Whilst I'd love to make something really big and loud, purely out of engineering interest, I'm sure the neighbours wouldn't.

So, a few bits and pieces were gathered together. I managed to score some unusual output valves for not a lot of cash from eBay seller seemoreitems. Some wire-ended N78 pentodes. I had a couple of mains transformers that I could lash up to get heater and HT from, and a rather nasty 8" 6 ohm Sansui "Woofer" from the 80's....

I was going to need to order a few things, namely a couple of chassis boxes, and an output transformer. I gave Philip at Bluebell Audio a ring, and bought a Hammond chassis and a 125D Push-pull output transformer.



So, gather the items together and see what transpires...

First up, the power supply. I orginally toyed with the idea of using two 6V transformers back-to back, but this was going to end up being huge. I found a nice 20VA 6V transformer, and a 50VA isolating transformer.. the junk box provided a bridge rectifier, and a couple of caps, all salvaged from a duff switched mode computer PSU

 Top side of the completed power supply....












And the rectifier and filter...

Note the tags on the top of those transformers have mains on one side, and our un-rectified HT on the other, we'll need to make a cover for that later...







Smoke test gives around 280V DC off load. Let's push on with the amp itself....

I'm going to start with the output stage and work my way back to the input, making it up as I go along!

First off is to look at the N78 datasheet, which can be found on the wonderful R-type.org website.

So, a pair in push-pull, class AB1, with 240V on the anodes (allowing for a little sag from our cheap PSU), should give us a nice healthy 5 watts or so, with a 9 kilohm anode to anode load. That's quite a lot for a small bottle, no wonder the N78 has a reputation for running hot.

So, how to mount our wire-ended valves?

I did contemplate mounting in a stainless steel cable gland, but it's a bit ugly, and I'm concerned the valve won't be able to radiate enough heat. I ended up drilling out some B7G bases, and feeding the wires though the holes and soldering them to the tags. Perfect. Be careful with the wires, if one breaks off, you've had it.

I mounted the output transformer, and started marking out for the valve bases.

I'll cut three for now, one each for the N78's and a B9A base for an ECC83 (or something) for the phase splitter.







Holes cut and tapped ....

Valves mounted....














Heaters wired up using twisted cable, it helps to keep the hum out. .... so it must be time for ....

The arty valve shot ...

One side of the ECC83 (12AX7) will perform phase-splitter duties, whilst the other will be our pre-amp.









After I got the phase splitter operating, I added another ECC83, this time as a cathode-follower to drive a tone-stack. I shamelessly pinched the design for the tone stack from Marshall, who stole it from Fender...

The other section is another gain stage...







I added some pots, a mains switch and the input jack... eventually it's my intention to mount the amp upside down, with the valves hanging down, so the layout seems wrong. Controls are (from left to right in this picture) "Drive", Bass , Mid , Treble, "Colour" and Gain.






Testing at this stage revealed it needed more "oomph" for that overdrive rock sound... Let's add another valve! ..  another ECC83.

Now things were sounding good... but left me with a problem, I had an unused triode section.

I thought about adding yet another gain stage, but that would just be silly, and probably noisy too... so I set about thinking about a tremolo.



I looked around the valve wizard website, which is a fantastic resource, and cobbled together a tremolo oscillator... I was hoping to couple the cathode of the oscillator to the cathode of the input gain stage to produce the effect. It didn't work...

What I needed was a method to modulate the volume... I had a cup of tea and a think. What about a small FET?



I had a look through the FET drawer, and selected a likely candidate,a BS107 in this case, but a 2N7000 would do nicely.) Connecting this temporarily to the volume pot worked well enough. I added depth and speed pots to the rear, as well as a foot switch plug, to switch the tremolo on and off.






Tidied up the wiring, and added a pink LED (because blue is sooo last year).













Now to tidy the workshop, and think about a cabinet....

Incidentally, this picture was taken with only the heater supply on, to show the valves lit up, and the LED working. Never, ever run a valve amp without a speaker or dummy load connected.






Here's the schematic...

Let's have a little walk through...

The signal enters at J1. C1 is used to block DC, as the grid of the valve is around 1.5V above ground, to set the bias of the valve. R1 is the grid leak, and also sets the input impedance. R4 is the grid stopper resistor, and, in conjunction with the miller capacitance of the valve and C4, stops the thing trying to oscillate or pick up radio. The signal is amplified by V1A, and emerges at the anode (plate if you're in the US). Anode load is provided by R8. The signal is coupled by C6 to the volume control, R10. The signal is then coupled to the drain of Q1, which is the tremolo modulator FET (more on this later..). The signal then passes via another coupling cap, C17 to another gain stage, V2A, in the same manner as the first. The output of the anode at V2A is directly coupled to the grid of the cathode follower V2A, which has no gain, but has a low output impedance to drive our tone stack. The tone stack consists of three filters, created by C11,12 & 13, R18 and the pots, which are used to control the amount of attenuation of each filter. The output of the tone stack is connected to the drive pot, coupled to V3A in a similar manner to before. The cathode of V1A has a pot connected, so we can change the bias of the valve slightly. This has the effect of altering any clipping that may be occurring, and, in a guitar amp, we want a bit of distortion! If the tube is biassed "cold" i.e. less current flowing (larger value of cathode resistor) the distortion has a "bluesy" sound to it, if the tube is biassed "hot" , it's more of a "rock" sound. The effect is subtle. The output at the anode of V3A is coupled by C17 to V3B, which forms the phase splitter or phase inverter. The grid of V3A is fixed at ~53V by the bias network, formed by R29 & R30. An inverted signal appears at the anode, and a non-inverted signal at the cathode. Note the anode and cathode resistors are the same value, ensuring balance (as long as the load is symmetrical). The inverted, and non-inverted signals are passed by their respective coupling capacitors, C18 & C19 to the grid of the N78 output valves. R33 (R34) provide grid leak, and R35 (R36) are the grid stoppers. Anode load is provided by the output transformer. R44 & R45 are feeding the screens. This resistor stops the screen grid from exceeding it's maximum dissipation during overdrive conditions. V1B forms a phase shift oscillator, and operates from about 2Hz to 10Hz, speed being controlled by R2. Each capacitor, which it's resistor, provides a phase shift of 60 degrees, and the inverted signal from the anode is inverted a further 180 degrees, forming the oscillator. The output from the anode is also coupled by a large value capacitor, C7, to R12 "depth" control which is used to feed the gate of the modulator FET, Q1. via the potential divider R13 & R14. The source of the fet is connected to ground, and the drain to our audio line, between C9 and C10. When the FET is biassed on, the audio is coupled to ground via the FET. Shorting the gate to ground switches the FET off, and switches off the tremolo effect.

How does it sound? Great... let's rock ...

.. now to learn to play the guitar !

Mobius HD Actioncam and RFI.

Now, being a gadget lover that I am, I decided to upgrade the rather old dashcam in my car to something a little more "HD". After doing a bit of research, and seeing the footage of a colleagues camera, I decided to get a Mobius 1080P Actioncam, purchased from a reputable source (beware the fakes). Good choice. Small unit, great pictures, sound is OK... except....

... it ruined my radio reception.





Now I have a newish car, and it's fitted with a DAB radio. After fitting the camera to my windscreen, I noticed DAB reception of my favourite station, Planet Rock, became patchy. Disconnecting the camera solved the issue.... hmmmm ... RFI ... Radio Frequency Interference.

So what was going on? I broke out my ancient spectrum analyser, and, equipped with a simple whip antenna, took some measurements....

Here's the DAB reception, centred on 220MHz, with the Mobius disconnected. The two humps are the two DAB Mux's (multiplexes) I can receive at home...










... and here's the same shot with the camera connected... yuck. I suspect the big peak to the right of the mux is effecting the AGC of the radio, and there's a lot of hash underneath it too, upsetting the signal to noise ratio... boo!








I tried a simple method of preventing RFI radiating from the cable of the device. I wrapped as many turns of the USB power lead round a random (ex-computer power supply) ferrite ring.









Brilliant....














... and after some field tests, reception restored!

Arduino Audio Wattmeter.

Many, many moons ago I needed something to measure audio power in a hurry. So I lashed up a quick and dirty audio power meter.

It was nasty....

... very nasty, but did the job at the time.

It was simply a switchable 4 or 8 ohm load (actually 4.7 or 8 ohms)  which was rectified, and used to drive two meters, each measuring full scale at 10 watts. The fan was driven from rectified audio. Most unpleasant.




Anyway, the load was good, as was the fan, but I needed some sort of better instrument, so a plan was dreamt up. Measure the peak voltage across the load, do a bit of maths and get the results displayed on a 1602 display, driven by an arduino. Great ....

So audio comes into the load (R1 & R2), and is rectified by a bridge. There's a small bit of smoothing on the output, so we can get a steady-ish DC. This is divided by a voltage divider, R3 and R4 and the resultant is fed to the arduino's A-D converter. Couple of things to note here, the load is switched, either 8.0 Ohms, or 4.7 Ohms, by S1. This also lets the arduino know which load is selected. S2 is a "BTL safety switch". It completely isolates the left and right channel. It also stops the arduino reading the right channel. If you're testing a BTL(bridge tied load) amp, or any type of amp which may not take kindly to commoning up of the negative poles of the speaker, best click this across. It also grounds the second analogue input to prevent spurious readings. A PWM signal is output from D6, and is used to drive a FET to run a fan. I chose a stupidly large FET here, you can pick something more sensible. I just happen to have a bucket of them. 

This will only give accurate results when feeding in a sine wave.

That's it really....

 Dummy load salvaged from the old unit...

Front panel coming together.

... sizing it all up....

... starting the wiring...

Looking good until I noticed that I'd mistakenly picked up a 16x1 display instead of a 16x2!... and the 16x2 requires a slightly smaller opening. Drat! ... some hot melt later will have to do (and it was all going so well ...)

Rectifier board and PWM FET

Rectifier was a little overkill! Still I had two identical, salvaged from a switched mode PSU.











Setting up ready for calibration...

Apply 28.28VDC accross the outputs of the two bridge rectifiers, and with the unit set to 8 Ohms, adjust each pot until the display reads 50.0W. Easy... (Why 28.28V ? 20 VRMS into 8ohms is 50Watts. 20*1.414 = 28.28Vpk... see here for more)





All calibrated and ready for action!
Top line on the display reads left channel, bottom reads right. The three readings are instantaneous power, average and peak.











And last, but by no means least... the code....

/* The "I'm sick of that analogue meter" Audio wattmeter (c) A.G.Doswell 2nd April 2016 License: The MIT License (See full license at the bottom of this file) V1.0 Arduino measures the DC output from a diode bridge across the load resistor, does the maths and displays the results. Peak-hold feature, averaging and PWM fan control too! This is only going to give true readings with a sine wave input. VRMS = VPk + 1.41 / 1.414 Power = VRMS^2 / R R is either 4 or 8 ohms (switchable) */ boolean rPin; int sensorValue; float voltagePk; float voltageRMS; float powerR; float powerL; float loadR; float peakR = 0.0; float peakL = 0.0; int loopCounter = 0; float averageL = 0; float averageR = 0; int averageCounter = 0; float averagePowerL = 0; float averagePowerR = 0; int fanPWM = 0; int fanPWM2 = 0; #include <LiquidCrystal.h> LiquidCrystal lcd(12, 11, 5, 4, 3, 2); // initialise the LCD void setup() { pinMode (7, INPUT_PULLUP); // switch connected to 7 defines load resistance, Open for ~8 ohms, short to GND for ~4 Ohms. lcd.begin (16, 2); } void loop() { rPin = digitalRead (7); if (rPin == false) { loadR = 4.7; // My 4 ohm load is actually 4.7 ohms } else { loadR = 8.0; // .. but my 8 ohm load is spot on! } sensorValue = analogRead(A0); // Output from Right channel connected to A0 voltagePk = sensorValue * (35.0 / 1023.0); //Convert the analogue reading to a voltage voltageRMS = (voltagePk) / 1.414; // Vpk to VRMS powerR = (sq(voltageRMS)) / loadR; // VRMS and load to power if (powerR > peakR) { // update peak value peakR = powerR; } averageR = averageR + powerR; //running average sensorValue = analogRead(A1); // Output from Left channel connected to A0 voltagePk = sensorValue * (35.0 / 1023.0); //Convert the analogue reading to a voltage voltageRMS = (voltagePk) / 1.414; // Vpk to VRMS powerL = (sq(voltageRMS)) / loadR; // VRMS and load to power if (powerL > peakL) { // update peak value peakL = powerL; } averageL = averageL + powerL; //running average fanPWM2 = (powerR + powerL) * 8; // get a value for the fan speed PWM output if (fanPWM2 > fanPWM) { // if it's greater than the current PWM value, then up date it fanPWM = fanPWM2; } if (fanPWM > 255) { // limit the maximum PWM value to 255 (flat out) fanPWM=255; } analogWrite (6, fanPWM); // write it to the if (averageCounter == 100) { // sets the Average reset averagePowerL = averageL / averageCounter; // create the two averages from the running value averagePowerR = averageR / averageCounter; averageCounter = 0; // reset the counter averageL = 0; // reset the running averages averageR = 0; fanPWM --; // decrement the PWM value for the fan if this are cooling down! } if (loopCounter == 1000) { // when this has run 1000 times reset the peak value. loopCounter = 0; peakL = 0; peakR = 0; } lcd.setCursor (0, 0); lcd.print (powerL, 1); lcd.print ("W "); lcd.print (averagePowerL, 1); lcd.print ("W "); lcd.print (peakL, 1); lcd.print ("W "); lcd.setCursor (0, 1); lcd.print (powerR, 1); lcd.print ("W "); lcd.print (averagePowerR, 1); lcd.print ("W "); lcd.print (peakR, 1); lcd.print ("W "); lcd.print (loadR); loopCounter++; // decrement the counters averageCounter ++; } /* Copyright (c) 2016 Andrew Doswell Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permiSerialion notice shall be included in all copies or substantial portions of the Software. Any commercial use is prohibited without prior arrangement. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESerial FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHOR(S) OR COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */

Pye 92 17" 'Transportable' Television set - 169 Chassis

Back in about 1981-82, I bought one of these sets from a jumble sale for 50p. I lugged it all the way home, stopping many times along the way (I was only 10 at the time). It didn't work. My mum kindly took it into the local TV shop (Seamen's Television in Carshalton) and they fixed it for me. It's a larger 17" TV, using a hybrid chassis of valves, transistors and 2 IC's, 625 line single standard, black and white.

It's more "trans-luggable" really, you can see that woman in the ad straining, it's got some weight!

The fun continued until about 1985 when the tube failed, curtailing the fun. It was binned.

I'd kept a casual eye on eBay for one for the last few years, and then in late 2014 one surfaced.

I collected it from the south coast and returned home with it.

Cosmetically in good nick, although the tuner was seized solid.

 A few repairs have been carried out in the past...

 .. but it looks OK.

Callins electrolytics are most likely to be faulty...

 TAA700 sync separator and video pre-amplifier chip.

These sets had a reputation for poor caps in the line stage, they've been changed in the past..

Mazda CME1713/ A144-120 tube. Flaky aqua-dag, but that shouldn't effect performance.









So, the mains filter capacitor was snipped out, and some mains applied via the variac.... and, the smell of burning dust and hot valves awaking from their slumber filled the workshop... a few moments later , a picture appeared! Tube looked to have plenty of life left. I went to the kitchen to get a cuppa, and on return was greeted with a VERY hot dropper resistor, and a line output valve glowing cherry red :(

Probably one of those rotten caps stopping the oscillator I thought ... I thought wrong. After an hour or so of testing, I pulled out the line transformer, and performed a ring test ...

No doubt it had developed shorted turns. The TV restorer's worst nightmare.

I asked everywhere to see if anyone had a replacement transformer. I had a few leads, but nothing was forthcoming....

It's probably the EHT overwind, I thought, so the plan was to remove the overwind and ring test that to see if that was at fault. I had some other unknown line output transformers I could take the overwind from.

Unfortunately nothing would shift the glue used to hold the winding to the ferrite core. Disaster struck, it broke up ...

I reassembled the set and stored it in the attic, until a suitable transformer could be obtained.

















Time passed, a lot of time......


and then, several weeks ago, a wonderful gentleman emailed me from one of the forums, who had seen my request for a transformer. He suggested a Philips 210/300 chassis transformer may be a suitable substitute. He had a Konig replacement, a ZTR371, which had the benefit of having a solid state EHT rectifier built in. It duly arrived in the post...

The set was removed from storage, and prepared for the transplant!

Both the Philips and Pye circuits were examined... there were differences. Mainly in how the line linearity is handled.

I initially decided to modify the Pye circuit, to accommodate the Philips positioning of the line linearity coil... stood well back and slowly increased the mains on the variac. Almost no width, but there's something....

I then changed the circuit back, and changed the wiring on the transformer to suit the Pye line linearity cicruit... results were much improved, although "boost" HT was woefully inadequate at about 320V, 850V would be better...






The reason for the lack of boost HT was quickly spotted.... another wiring change...

It's all a bit of a lash up of bits of wire and croc clips, but it IS working.

During this time I noticed there was now a precious lack of "snow" on the screen. I decided to stop here, and evict those awful Callins electrolytics, and anything else which looked suspect.

 ... and un-seize the tuner.

 ... lots of surface rust on the tuning spindles, which is cleaned off using a little WD40 on a cotton wool bud.

The cam-gear on the other side is also stiff, but eases with a little WD40.

After it's all moving again, I cleaned off the WD40, just in case it attacks the plastic nuts on the spindles. A good coating of spray grease should keep everything lubricated.








So, the tuner is reassembled and fitted back to the set. I can now tune in the set, and can *just* see a picture, but it's totally lacking in contrast. Time to evict all those nasty electrolytics. Many are either physically leaking, cracked or electrically leaking or low in value. Sadly, this doesn't fix the issue. On a brighter note, the height and width are improving, probably due to the valves waking up after a long slumber! I add a 100pF capacitor across the line harmonic tuning cap, which improves the amount of adjustment available on the width control.

I mount the line output transformer permanently, and tidy up the wiring.

So, back to the lack of contrast. The video is demodulated by the TAA700 IC. Expect for in this case, there was precious little video coming from the IC. I find a few resistors outside of spec, but changing these makes no difference. I fit a new IC. No change.

Scoping up some of the waveforms around the IC shows it not quite getting the right information back from the line output stage to correctly operate. Unsuprising considering the wrong line output transformer is fitted. It's probably stuck in blanking...

I turn my attention to the line stage, and R 92 in particular.

I connect a 1 megohm pot and a 100K resistor in series, and connect it across R92, in an attempt to increase the drive to the IC.









Tune for maximum smoke. Only it doesn't work, as I get the required pulses at the IC , the picture gets worse! ... So I re-wire the pot so as to decrease the pulses to the IC!

Bingo! Pots of contrast. I disconnect the pot and take a measurement of the value. It's near enough to 1 megohm, so R92 is replaced with a fixed resistor.

Now just to centre up the picture and soak-test it...






... if only...

You might glaze over at this bit ....

After being on for about half an hour, the picture faded out rapidly. I cut the power quickly. Is this the return of the fault that had killed the original transformer??? The new transformer is running a little hot. Perhaps a bit hotter than I'd like. The boost capacitor, and the first anode decoupling capacitor have both been replaced in the past, it was probably in the 80's at the latest, so I order some up. Replacing these makes no difference, and half an hour later the set cuts out abruptly again, as the line output stage shuts down. I wait for a few minutes, and scope up the line drive signal to the PL504 output valve. It remains strong, even in the fault condition, so something is loading down the line output stage. I change the PL504 and PY88 boost rectifier, and I'm rewarded by more width, so I remove the 100pF cap I'd previously fitted to improve the width. Sadly after a few mins, the line stage once again shuts down. I move the width stabilisation feedback from the direct tap on the transformer, to the capacitively coupled tap. This improves the range of the width control further, but doesn't sort the fault... phut! The set goes off.... I give any remaining suspect components in the line output stage a liberal squirting of freezer (on the cheap).  Sadly it appears the line output transformer can't cut the mustard. It's failing when warm. I'm about to give up. It has taken hours and hours of work to get this far, and I hate to be beaten, but enough is enough.

I made some tea, and studied the circuit diagram again. Nothing is really forthcoming. I decide to cut apart the original Pye transformer to see if I can learn anything from the windings. I make some simple resistance measurements and compare them to the Philips transformer. The winding I'm supplying the HT to is a lot lower resistance than the Pye transformer. It should therefore be drawing more current. I scope up the current flowing into the transformer using a totally inappropriate hall-effect amp-clamp, which appears to tell me there's 2A flowing into the lopt. I suspect there's a lot less than this in reality. I can see there's an unused tap on the primary of the Philips transformer, so I hook up the HT feed to this tap. Now my (lying) current probe tells me there's an amp flowing into the transformer, well that's a change in the right direction. The picture comes up, and stabilises, the width is still good, and stays there .... for hours .... Success at last!

So for future reference, here's a quick sketch on how to fit a Philips 300 transformer to a Pye 169!

Some of the guilty & redundant parties...












Obligatory arty valve shots (seem to be getting the hang of this Nikon camera now!)

...and here's a video of the thing in action (maybe not so good with the new camera!)...