MTX Power Supply Details

There are an increasing number of MTX computers appearing for sale on ebay without power supplies, as you are unlikely to be able to find an orphan PSU, alternative methods of powering an MTX computer would need to be considered. To that end, a description of the standard MTX power supply should be helpful.

Background

The Memotech manuals do not give any details on the MTX external power supply, the only published information coming from the label on the top of the PSU, most PSUs that have been seen are marked "Output 22.5 VAC 1A Tapped at 18V and 9V". The PSU with Keith Clatworthy's low serial number MTX512 has additional information - "Output 22.5 V ~ 1A. 18V ~ 0.82A. 9V ~ 0.28A.

Inspection of the PSU internals has confirmed that the "PSU" is in fact only a multi-tapped transformer, all voltage regulation & smoothing is done on the MTX computer board as shown in the schematic below.

Based on this information, it can be seen that there are two main options available to provide alternative power for an MTX :-

• Obtain a like-for-like replacement for the original PSU

  • Providing AC voltage outputs and requiring no modifications to the MTX computer board

  • Totally interchangeable between different MTX computers

• Provide an alternative method of providing the DC voltages required by the MTX system board

  • Modifying the MTX computer board as necessary to accept DC inputs of +12V, +5V and -5V (-12V optional).

  • Making the modified MTX only suitable for use with the redesigned PSU

Original MTX Power Supply

MTX PSU Connection Details (When viewed looking into the plug on the end of the PSU lead)

The secondary of the PSU transformer is centre tapped at 9V - 0V - 9V, with an additional tapping at one end of 4.5V, leading to the nominal voltage described on the PSU of 22.5 VAC, the other taps give nominal voltages of 9V between pins 5 and 3&4 and 18V between pin 5 and pin 2. The MTX Service manual advises that the voltage at J9 on the computer board should be between 23V and 24.5V - this value is with the PSU under load, the voltage measured with the PSU disconnected from the MTX, i.e., with no applied load, can be expected to be somewhat higher.

AC voltages are usually described and measured in terms of the Root Mean Square (RMS) voltage, in simple terms, the RMS voltage of an AC voltage is the equivalent DC voltage that can produce the same power output. For a pure sinusoidal waveform, the RMS voltage is 0.707 * the peak voltage. Thus, at the 9V output of the secondary winding in the MTX transformer, the peak voltage is approximately 12.7 V and the peak-to-peak voltage approximately 25.5V.

Therefore, the magnitude and phase of the MTX computer board PSU inputs would be similar to those shown below:

 

MTX Voltage Regulation

This is the circuit diagram from the MTX Operator's Manual, although Tony Brewer has identified that the values shown for C51 and C53 are incorrect and the values from the Service Manual are 4700uF 16V (not 1000uF 25V) for C51 and 100uF 25V (not 470uF 16V) for C53. Similarly, C38, C39 and C40, shown as 1uF have been replaced with 10uF 16V radial electrolytic capacitors. These values agree with the components seen fitted to a number of MTX computer boards.

Simplistically, full wave rectification can be thought of as inverting the negative half of the AC sine wave. The resultant voltage would be a pulsed direct current as shown, with the amplitude varying between 0 and approximately the peak voltage of the input.
Capacitors are installed across the outputs of the full wave rectifier to provide a smoothed DC output. Capacitor values are chosen to meet a specified tolerance specification for "ripple" - the small variation that you can see between the peaks in this diagram.

 

+5 Volts DC

The highest current demand in the MTX is on the +5 VDC supply which is supplied from the full wave rectifier comprising of diodes D15 and D17.

On the positive half of the mains ac cycle, current flows from the upper 9V winding tapping via connection J9/2 and diode D15 to capacitors C55 and C56, which both charge to the peak value of the ac waveform. The "return" current flows from the negative terminals of capacitors C55 and C56 to the 0V tapping via connections J9/3 and J9/4.

On the negative half of the mains ac cycle, current flows from the lower 9V winding tapping via connection J9/5 and diode D17 to capacitors C55 and C56, which both charge back up to the peak value of the ac waveform. Again a "return" current flows from the negative terminals of capacitors C55 and C56 to the 0V tapping via connections J9/3 and J9/4. The action of the positive and negative currents combine to increase the available current.

The "smoothed" voltage of about 12 VDC is fed to REG2, an LM7805 voltage regulator. However, an LM7805 can only produce a stable output current of about 1A so additional power is required to run the MTX. The TIP2955 (Q4) is a PNP power transistor used to augment the output of the LM7805. The TIP2955 would allow the 5V circuit to supply about 5A, well in excess of what the MTX requires, so fault current protection is provided by FS1, a 3.15A fuse.

With little current flowing through the LM7805, there is a small voltage differential between the emitter and base of Q4. As the power through the LM7805 increases, a larger voltage differential between the emitter and base of the TIP2955 is created until it reaches about 0.65 to 0.7V, when the TIP "turns on", allowing current to flow through from the emitter to the collector, to the 5V line, supplementing the output of the LM7805.

The wire-wound sense resistor, R62, is used to bias the current between Q4 and REG2, using a value of 10R, the current provided by the regulator is very low (<100mA), the majority of the current is switched through Q4. 

-5 Volts DC

The -5 VDC supply is supplied by a another full wave rectifier (D16 and D18), with the diode polarity reversed as a negative voltage is required. The Zener diode (ZD3, a 5.1V type) and resistor R60 form part of a simple type of shunt regulator. The Zener diode is manufactured to conduct current at its rated voltage when reverse biased. As long as the current to it is limited (which is the function performed by resistor R60) it will hold the -5V line at -5.1V.

The +12 volt circuit is interesting, as drawn, the circuit diagram suggests that the +12 VDC supply was intended to be supplied by the full wave rectifier comprising D14 and D19. However, Mark Kinsey has pointed out that in reality, the circuit will function as a half wave rectifier. D14 acts as a half-wave rectifier for the winding connected to J9/1, this will charge capacitors C51 and C52 to 19.1V dc [ 13.5 x 1.414 ]. Compare that to the peak voltage from J9/5 via D19, which is 12.7V [ 9 x 1.414 ] As long as the voltage on capacitors C51 and C52 is still > 12V, diode D19 will not conduct as it is reverse biased.

Comparing the expected voltage trend in this scenario with the example above, you can see that with no contribution from the negative half of the mains cycle through D19, the smoothing capacitor would have to discharge for longer, resulting in increased ripple on the filtered DC output. This is likely to be the reason why C51 was replaced with a larger value capacitor.

This also raises a question on the revised voltage specification for C51, it is clear why the capacitance value was increased to 4700uF from 1000uF, but given that the selected voltage for an electrolytic capacitor should always exceed its peak operating voltage, a reduction in the selected capacitor voltage from 25V to 16V is not logical.

The "smoothed" DC supply from D14 (& D19) is fed to REG1, an LM7812, referenced to the common rail for the DC side of the circuit - which is also at the same potential as the centre tap of the transformer. The differential voltage input to REG1 appears to be about 18 VDC, allowing REG1 to provide the regulated +12 VDC output, but is above the rated voltage of C51!

Observed Transformer Output Voltages*
Although three different styles of MTX PSU have been seen, apart from another, possibly pre-production, unit, all of the transformers seem to have very similar voltage outputs. I do not think that the transformers used in the three models of PSU were actually different. If  you measure your own PSU voltages, let me know and I will add them to the table below
Measurement Points - between PSU plug pins (no load connected) (When viewed looking into the plug on the end of the PSU lead)   (All voltages RMS AC)
	1 (Nominal 13.5 VAC)		2 (Nominal 9 VAC)		3  (Nominal 9 VAC)		4 (Nominal 22.5 VAC)
PSU	VAC	% Var.+	VAC	% Var.+	VAC	% Var.+	VAC	% Var.+
512 (1)	14.1	-8.4	10.1	-2.6	9.6	-4.0	23.7	-7.2
MCL (2)	15.4	+0.1	10.0	-3.6	9.5	-5.0	24.9	-2.5
No Label (3)	19.1	+24.1	11.7	+12.8	11.7	+17.0	30.8	+20.6
512 (ds1)	14.18	-7.9	10.18	-1.8	9.68	-3.2	23.86	-6.6
512 (ds2)	15.78	+2.5	10.68	+3.0	10.65	+6.5	26.43	+3.5
512 (ds3)	16.45	+6.9	11.19	+7.9	11.16	+11.6	27.61	+8.1
FDX (ds4)	15.55	+1.0	10.1	-2.6	9.6	-4.0	26.15	+2.4
512S2 (ds5)	16.26	+5.7	10.33	-0.4	9.83	-1.7	26.09	+2.2
X	15.4		10.4		10.0		25.5
Max.	16.5		11.2		11.2		27.6
Min.	14.1		10.0		9.5		23.7
σ	0.86		0.39		0.59		1.33
The mean values shown are based on averaging the voltages of the "typical" transformers, i.e., rejecting the suspect values in row 3. + The % variation column shows the variation of the individual transformers from the mean. The min, max and standard deviation values are calculated on the same basis.
Click on the camera icon to open a photo of the style of PSU being measured (not the actual PSU)
As noted earlier, an off load transformer will produce higher secondary voltages compared to a transformer at full load. This is known as transformer regulation (there is a good explanation of this on  www.allaboutcircuits.com. For small transformers this can be between 10 to 20%, so taking an average of 15%, the measured voltages with the transformer off load can be expected to be up to the values shown below.
+15%	15.5 (13.5)		10.4 (9.0)		10.4 (9.0)		25.9 (22.5)
These values are remarkably close to the average values obtained from the, admittedly small, data set.

*The output voltages of the MTX PSU are also subject to variation in the mains voltage. In the UK, the voltage specification for domestic supplies is 230 VAC +10%, -6% - this was originally 240 VAC +/- 6%, but was modified to be in line with the harmonised voltage specification across Europe, without actually changing anything. I have found data on the web that suggests that the average voltage is still 240 VAC - the specification in place when the PSU was manufactured, although the service manual shows that the transformer installed was a 220 VAC unit.

 LTSpice IV is available free from Linear Technology

Curiouser and curiouser!
The notes on this page are believed to provide an accurate description of the MTX power supply and applies to all models of the MTX. However, some PSUs supplied to the European market are fitted with an internal fuse on the low voltage side of the transformer which makes no sense to anyone who has reviewed the MTX power supply design. If you can shed any light on why it might have been fitted, please let me know.
This photo is of the internals of a UK spec MTX power supply. At the top you can see the mains cable, DPST mains power switch and its connections to the primary side of the transformer. At the bottom, you can see the connections to the cable for the MTX 5-pin DIN power plug.
This photo, courtesy of Steven, shows the internals of a MTX PSU with a European mains power plug. As you can see, there is an internal fuse connected to the centre tap on the LV side of the transformer which feeds the 0V line to the MTX. This fuse has been seen on more than one PSU so it was apparently fitted by Memotech but the reason is unknown. In fact, should it blow, rather than offering any protection to the MTX, it could result in serious damage to the machine, particularly the VRAMs.
Why it doesn't make sense . . .
As most people probably know, UK mains plugs are always fitted with a fuse between the live socket terminal and the appliance. This fuse is intended to protect the cable, NOT, the appliance. The ring final circuit design that's still popular in the UK provides for very high currents to be supplied to the connected appliances. Should a fault develop in the appliance, the fuse is intended to blow before an overcurrent situation could damage the cable, potentially leading to a fire. (Before appliances were fitted with molded plugs with manufacturer fitted fuses, new plugs bought from retailers for customer fitting were often supplied with 13A fuses which, in many cases, were never replaced with fuses more appropriate to the cable size. Hopefully, you don't have a 13A fuse in your MTX power cable! Molded plugs were introduced to mitigate that risk.) The two pin plugs used in Europe, where ring final circuits like we have in the UK are not used, are commonly not fused as the protection fitted to final circuits at the distribution board allows for lower maximum fault currents. When I first saw the fuse in a MTX power supply, I thought that it might have been there to provide similar mains cable protection as the UK plug-top fuse. However, on closer inspection, it can be seen that the fuse holder is fitted to the low voltage side of the transformer and is in fact installed between the centre tap on the LV side and wires that connect to Pins 3 & 4 of the MTX low voltage DIN connector. The fuse provides no cable protection. So, what does it do? I have absolutely no idea! What is certain is that if the fuse were to blow, there is likely to be an adverse effect of the -V line which feeds the Zener diode (ZD3) used to generate the -5 VDC level in the MTX. This would almost certainly result in damage to the 4116 VRAMs which are very sensitive to out of spec voltage differentials between their three supply voltage levels (+12V, +5v and -5V) and a rated maximum of -5.5v between the -5v pin (Vbb) and 0v (Vss).