Friday, 23 October 2015

Radford SC2 Control unit & pre-amplifier.

Remember the nice Radford STA15 I did a couple of months ago? (You can check it out here) Well, this is it's matching control unit & pre-amp.

The four ECC83s (12AX7), all check out well within limits, and still nicely matched after years of service. Those main smoothing caps also still performing well, and needed very little in the way of reforming.
Very little wrong with this , except it was full of those Wima caps that tend to short circuit without warning. They'll have to go. There's also a few 50uF 6V electrolytics that are also in poor shape. Noisy switches and pots are taken care of with a little cleaning.

 It's owner was saying that the recording output lacked bass. Indeed, mainly due to one cathode bypass being open circuit, and the other reading 7uF of it's original 50uF. Note the brown discolouring round the end cap. Let's get to work....

Replacement caps...

... and finally re-united and playing nicely together once more.

Tuesday, 20 October 2015

Weller soldering iron add-on.

Now I'm a great fan of the Weller TCP "Magnastat" iron; robust, simple, great parts availability. Yes, there are better irons out there, but these are usually available secondhand for peanuts.

They are available in various wattages, and I have two, both 50W, one with a nice fat screwdriver type tip, for valve stuff and tagstrip etc, and a slender one, for more delicate work.

The tip has a temperature rating stamped on it's base. This contains a special magnet, which loses it's magnetism at a certain temperature (known as Curie temperature). This magnetism is used to operate a magnetic switch (like a reed switch, but a little more robust) to switch the element on and off, which heats the tip.

As bench space is limited, I only use one power supply, and got sick and tired of pulling one iron plug out of the power supply, and plugging the other in when I wanted to switch between irons. What was needed was a switch box....

... then I got a bit geeky (well, there's a surprise!). Let's have an led on the box, to show when the iron is heating. Useful to know if the iron is up to temperature, or if a fault has occurred (occasionally the switch or element fail , see here ) , so I devised a simple circuit to put an LED on when the iron is drawing current.

Simple. When the iron is drawing current, there is a small voltage dropped across R1, which causes the op-amp IC1A to attempt to drive it's output to rail. Now the current flowing in R1 is obviously AC, so the output is a 50Hz (or 60Hz, depending on your mains) waveform, but it's enough to light a red LED. The power supply for the op-amp is a crude supply derived from the 24VAC from the weller PSU. It's half-wave rectified by D1, and D3 and R2 Clamp the voltage peaks to 5.1V. It's smoothed by C1.

It's crude, horrible but works!

Monday, 19 October 2015

Car subwoofer filter.

This is a bit of a random one, left over from the start of the year, but I thought I'd put it up anyway.

I got a new car earlier in the year (well, new to me!). It has a very nice radio in it, and the speakers aren't bad. It has DAB, USB , Bluetooth, and, after a long while looking for the slot, no CD! It also lacks a line output or remote line to switch on my after market sub-woofer and amplifier, but it does have an unused speaker output, so I knocked up this:

Now the issue is, the output from most head units is bridge-tied load, to get a bit more power out. This leaves the speaker lines floating at half supply (6V) to chassis. Never ground a speaker wire!

This is OK though, as we can use this to trigger our remote line.

One side of our speaker is connected via R10, to the gate of a small FET (Q1), this, in turn switches on Q2, which supplies 12V to our remote line, to trigger our amplifier on.

IC1A gives us a low-current half-rail power supply, we can use to reference our audio to, as we can only "swing" our audio above the chassis 0V, rather than swinging both ways, as we don't have a negative supply rail.

The speaker is also connected to a pad, formed by R4 and R9, which lowers the audio voltage to something more suitable to drive our amplifier. IC1B is simply a unity gain non-inverting buffer amplifier, and this feeds an adjustable low-pass filter, formed by the dual-gang pot R14, R15, C7, C8 and C9 around IC1C. R15 is an output level adjust, and finally IC1D is the output amplifier, with a gain of 2 to make up for the loss in the filter. R19 is there just to limit current. C10 blocks the DC content of the output (remember it's swinging around 6V, rather than chassis) and R20 references the audio back to the chassis.

S1 is used to switch the phase if the sub makes the mid-bass sound incorrect. R11 is a 10 ohm load resistor. Some head unit amplifiers may not like being run unloaded, so it's there to give some load. You might not need it, and it will certainly do no harm to the head unit if it's not there. Add it if you need to, but make sure it's well rated to dissipate the required power.

Thursday, 15 October 2015

Vintage test gear - Heathkit Harmonic Distortion Analyzer IM-5258

A quick description of the Heathkit Harmonic Distortion Analyzer (Analyser!)

Great bit of vintage 70's 80's test gear.

I've tried a couple of times in my life to build a THD analyser, before I gave in and just bought one... I think I'd be OK today, but I have a reliable instrument, so it saves me the grief!

It takes the input signal, a sine wave, and buffers it, matches it for voltage with it's own internal oscillator, then notches out the input signal. What then remains is the harmonic distortion and noise. Simple to say, a little more complex to put into practice.

The Q-factor of the notch filter must be high enough so it doesn't notch out the 2nd harmonic (or 3rd, 4th, 5th etc) , which at audio frequencies can be difficult to achieve. The resultant output needs to be measured with a high bandwidth AC RMS meter, sensitive to microvolts, which can also be a tall order!

It's also tricky to keep the whole thing from oscillating.

Thankfully, I picked up this:

It's all discrete, not an op-amp in sight, and just works! It's also auto-tuning, but you still need to set the range and adjust the notch, so I'm not sure how "auto" it really is!

So, when we're designing an oscillator or amplifier, we can check how much distortion it really is producing, and where in the frequency spectrum it is! Useful stuff....

Here's a quick video of the thing in action...

Saturday, 10 October 2015

Arduino Mains Monitor with SIM900A GSM messaging.

Picture the scene... You're away on holiday, sunning yourself in Spain (or if you're Spanish, enjoying the damp weather in the UK !) ... and thousands of miles away, some spurious minor electrical niggle in your house causes the RCD circuit breaker to open. You return 10 days later to find the contents of your freezer oozing their way across the kitchen floor. Not nice... You reset the breaker, and the power comes back on, showing no faults.

It happens from time to time in my house. I've tested each circuit, and every appliance to try and find the cause of the random tripping. The fault is not visible on any of my insulation readings, but, nevertheless it does happen. Sometimes not for years.... frustrating!

So what can we do about it? Fix the fault would be the easiest thing to do, but it's eluding me. How about monitoring the mains, and sending me a message, so if it does trip, I can reset the power without issue? Good plan...

Now the disclaimer:

WARNING. Do NOT try this at home. This project deals with mains voltages. A shock from the mains will hurt and can easily be fatal. Work safely. Use an RCD. Disconnect from the mains before making any adjustments. 

I will not, under any circumstances, accept any liability if you decide to re-create this project for yourself.

Now that's got that out of the way...

So, we have a plan. Measure the mains voltage (and why not frequency at the same time?), and act conditionally on it's failure. We also need to get it to send us a message. We're going to need some form of uninterruptable power supply, so when the mains does go off, our micro will still be running, and we've got power enough to send the message. Here's the circuit:

I'll go through it step by step...

Power comes in via a fuse to two transformers. TR1, is a mains to 15V transformer, and is used as our power supply. It feeds a bridge rectifier, B1 and C1 and C2 are used to smooth and decouple the resultant DC. The DC is fed via D2 to a 7805 regulator, which provides the 5V for the Arduino and the SIM900  module. So that's fairly straightforward. The 7805 in this instance is a 2A part, as the SIM900A does require a fair amount of current to function.

(You could, and it would be safer to do so, build this unit using two wall warts, one, with, say at 15VDC, 3A output to replace TR1 & B1, and a second, with AC output in place of TR2. That way, all the tricky and potentially dangerous stuff with the mains is eliminated)

The rectified DC from B1 is also fed to two LM317 voltage regulators in series. The first, IC1, is configured as a 100mA constant current source, this is then used to feed the second regulator, IC2, which is configured as a conventional voltage regulator, in this case, R3 is adjusted to provide around 14V. This forms a 100mA constant current charger, with maximum voltage of 14V, which we can use to charge G1, which is an old 12V NiMH battery I happened to have kicking about. It's got plenty of capacity left, so will do nicely as our back up battery. S3 is a battery diconnect switch. Useful for resetting. (Why didn't I use an LM200, as it is capable of both current and voltage regulation in one package, instead of two LM317's? Because I didn't have one!)

So, when the mains is present, our battery is being charged. In the event of mains failure, D2 will stop conducting, and the supply will be seamlessly taken over by the battery, supplying current to the 7805 via D1. D3 prevents current flowing back into the charging circuit. The battery voltage is also sampled at the mid-point of the potential divider, R11 and R12, and fed to the Arduino A1 pin. T2, between the reference pin of IC2 and ground, is used to switch off the charging circuit momentarily, so we can measure the battery voltage, and not just the output of the charger. It's controlled by the Arduino A2 pin, configured as a digital output.

OK, so that's power and back up power sorted, so what about measuring the mains? TR2 is a small (3VA) 18V mains transformer. This feeds another bridge rectifier, B2, and a small smoothing network, formed by C7. The voltage developed across C7 is fed to a potential divider formed by R5 and R4, and the resultant voltage fed to Arduino A0. This voltage will be directly proportional to the mains input voltage on the primary of T2.... or will it? Whilst I was experimenting, I noticed the measurement wasn't linear. Adding a bit of loading, in the form of R15 to the secondary helped matters no end, although it was accurate enough over the range required for this to be left out, if required.

Also coupled to the secondary is our frequency measuring network. The AC is fed via coupling capacitor C8 to the base of a BC547, T1. This provides a 5 volt 50 Hz signal to the input of IC4A (a Schmitt trigger hex interter), this will square our pulses up. The output of IC4a is fed via a low pass filter, formed by R8 and C9 to the input of IC4B, the output of which is connected to the Arduino pin D8. We'll use this to measure the frequency of the mains. Tie all of the unused inputs of IC4 to ground.

On to the micro side of things...

The LCD is wired to the Arduino in the time-honoured fashion, except for the cathode of the backlight, which isn't connected straight to ground, but to the collector of T3. This allows us to control the backlight, using A4 of the Arduino configured as a digital output. There's a momentary switch coupled to A3, which is used as a "push to transmit" function, and a toggle switch, S2, connected to pin A5, which is used to stop the SIM900 sending messages. R9 is the LCD's contrast control.

The SIMR pin on the SIM900 module is connected to the Arduino's hardware serial Tx pin, and the SIMT pin is connected to the Rx pin.

The software.....

We're going to need something to measure our frequency. I did initially use PulseIn , but it's not that accurate, so I switched to using the most excellent FreqMeasure library, available from . Now this has a drawback. It uses Int 1, and this is in conflict with the software serial library, which is why the SIM900 is connected to the hardware serial port of the Arduino. The issue here is the hardware port is also where the inbuilt USB interface sends/receives data, so you'll need to disconnect the SIM900 from the arduino whilst uploading the sketch, or doing any serial debugging. You can use SoftwareSerial during development, but be prepared for some unusual responses from FreqMeasure!

Ok, so the sketch.

There's a rake of variables set at the start. The variables to watch are the mains tolerances, these will need to be set for your local mains (They're currently set for UK mains spec.)

  float MainsMinV = 216.2; // This sets the lower limit for the mains voltage. Change this to suit your local voltage limit
  float MainsMaxV= 253; // Maximum voltage limit
  float BatteryMin=11.2; // Battery low limit
  float MainsMinF=49.5; // Minimum allowable mains frequency limit, change to suit local power
  float MainsMaxF=50.5; // Maximum allowable mains frequency limit.

You will also need to change line 260 :
      Serial.println("AT + CMGS = \"+44xxxxxxxxxx\"");// recipient's mobile number, in international format
If your mains is not ~240v, you will also need to change the scaling factor in the software , line 286, so the voltage reads correctly. Currently the mains voltage is scaled so 250V is equal to 5V at our micro. If, say, you're on 120V mains, and want the voltage to top out at , say 130V, then the scaling will be 130/5 = 26. It's currently set to 51, so 5V on our analogue port reads as 255V :
  ACVoltage = (sensorValue * (5.0 / 1023.0))*51;

Adjust R4 to get the mains calibration correct.

So, here's the complete sketch:

If you're using a SIM900A module outside of Asia, you may run into difficulties, as I did. There's some brief notes I made here : and a link to a website which contains detailed instructions on sorting out the firmware to make the unit work.
Here's some pictures of my unit:

BTW. Sainsbury's is my service provider!

Electrical Safety is paramount. Here is the incoming mains earth (ground) securely tied to the chassis of my metal case. 

Friday, 2 October 2015

SIM900 (and SIM900A) module signal strength.

I've been working on a mains monitor project for a while, and I've wanted to interface it to a mobile phone module, so it will send me data. I'll publish the finished thing up soon, but , in the meantime, reading forums etc, people seem to have a couple of problems.

1) Getting the oh-so-very-cheap SIM900A module to work.

2) Getting a signal strength report from the thing.

The first problem appears to be that the SIM900A module, which I purchased for less than a tenner from eBay, doesn't work outside Asia. Yep. It certainly doesn't work here in the UK. But, never fear, there's a website with all the info to help us out. which, after a little fiddling about, had provided a firmware cure. Now I have used my little FTDI module to flash the firmware on my unit at 57,600 Baud. The website claims you need to set the baudrate of your device to 460800 baud, which mine won't support. It worked fine at 57,600, however. I did have to wait a little while (about 20 mins) for the process to complete though.
The second problem seems to bug a lot of users. I searched in vain for a ready-to-go library to solve all my woes, but couldn't find one, so here's a little routine to get the signal strength out of the unit, so you can do with it what you will...
#include <SoftwareSerial.h>   
  SoftwareSerial SIM900 (6,7); // SIM900 connected to pins 6 & 7 (pin 6 to SIMT and pin 7 to SIMR)
  char Reply[200]; // Serial buffer length
  int number = 43; // msg length
  int Signal; // Signal strength as reported
  int temp1; // 4 temporary integers, used to extract the data from the incoming serial.
  int temp2;   
  int temp3;   
  int temp4;   
  int BER; // Bit error rate
  int SignaldBm; //Signal in dBm
void setup() {   SIM900.
  SIM900.begin(9600); // Start serial comms with module
  Serial.begin(57600); // Start serial comms with PC
void GetStatus() {     SIM900.
    SIM900.write("AT+CMGF=1\r"); //set GSM to text mode
delay (150);   SIM900.
  SIM900.write("AT+CSQ\r"); //Send command for signal report
  delay (200);   
  while(SIM900.available() >0 ) { //wait for responce
    for(int i=0; i<44; i++) //read character into array from serial port
    Reply[i] =;   }  Reply[199] = 
 Reply[199] = '/0';       temp1=Reply[31]-
  temp1=Reply[31]-'0'; //convert relevant characters from array into integers
  temp2=Reply[32]-'0';   temp3=Reply[33]-
  temp3=Reply[33]-'0';   temp4=Reply[34]-
  if (temp1 == -48) { //if temp1 is -48 then do it again as the data is not valid (yet)
  if (temp3 == -4) { // use temp3 to determine where in the array the relevent integers are (-4 is a ",")
    Signal= temp2+ temp1*10; // calculate signal if the first digit is a multiple of 10
    BER = temp4;
  else{    Signal= temp1; 
   Signal= temp1; //calculate signal if the first digit is not a multiple of 10
   BER = temp3;
  if ( Signal == 99) { // if our signal is 99, IE no signal condition , the return a signal of -1
    Signal = -1;
  SignaldBm = Signal*2 -113; // calculate dBm for geeks like me.
  Serial.print ("Signal: "); //output stats to serial interface.
  Serial.println (Signal);   
  Serial.print (SignaldBm);   
  Serial.println ("dBm");   
  Serial.print ("BER: ");   
  Serial.println (BER);    }
void loop() { GetStatus ();  }
GetStatus (); 
There , I hope that eases the pain for a few users! Have fun, and check back soon for the mains monitor!