Embedded micro's start up procedure.

[samjmann]

I've been working on a early 90"s tape deck. It has a 64pin flat pack micro controller. The problem is it is not running at all.

It has: 5v Vdd, Reset, no 6mhz xtal osc, also a mains on detection to the micro.

I know what you are thinking... it won't work without the xtal running. Am I right in thinking the xtal should be running as soon as Vdd is applied, then the reset is set and the micro runs?

The ac mains detect line is correct, the waveform on the pin is very similar to the manual.

Not trying to be product specific here, more first principles of operation!!

Any thoughts anyone

[duncanlowe]

Be careful when testing the OSC pin. The load of a scope probe can actually stop the oscillation.

Often startup sequence would be: 5v is apllied, but a timing circuit (usually R/C) holds the device in reset to allow the power to be stable. Sometimes the reset is controlled by something cleverer, for example a voltage supervisor that only releases reset when the supply is stable. I'd expect the OSC too be up and running even when in reset, but I don't know for sure.

[Maarten]

It might start upon reset, probably also depends on the type of controller (a datasheet will probably have this information) hence the often implied product specificity when asking a question related to a repair ;-) In this case the most relevant product specificity would be the type and brand of controller used.

[cmjones01]

A micro of that vintage should *probably* start the crystal oscillator at powerup, as long as it's not being held in reset, since early 90s microcontrollers didn't tend to have internal oscillators to get them going. However, these things are very device-specific, as others have said. The good news, though, is that a crystal oscillator of that era should be plenty powerful enough not to be disturbed by a 10:1 scope probe, so you should be able to see the oscillation.

Chris

[SiriusHardware]

Can you be more specific about the micro in question (markings, any ID numbers on the circuit diagram if you have one).

Normally, crystal oscillators with such relatively low frequencies will battle on if you connect a scope probe to them, although the frequency of oscillation may 'bend' slightly for the duration. I tend to find it is the ones which are in the tens of MHz which are prone to being stopped by scope probes. If in doubt, if you have a receiver which can cover 6MHz, put it into SSB or CW mode and dangle a pickup aerial near the oscillator. With the receiver tuned very near to the frequency of the oscillator you should be able to hear a steady tone equal to the audio frequency difference between the oscillator frequency and the frequency to which the receiver is tuned. When you switch the unit off the tone should disappear, when you switch it back on it should reappear.

If you don't have another 6MHz crystal to hand, try any crystal which you do have which is near-enough. If that oscillates it will be enough to tell you there is a problem with the original crystal - of course any timing or other operation which is derived from the original 6MHz will be a bit off, but it will tell you whether you need a new crystal. 6.00MHz is a fairly common uP crystal frequency especially for the 8051 family series and derivatives.

[G6ONEDave]

If you have the circuit, check that the micro has all of it's supplies, I've found some that have more than one. Likewise osc, reset and chip select all need to be present where used. Another thought is if it uses a memory of some description and is that OK and not hanging due to a bad back up battery or capacitor. Other possibilities are down to if the micro does a test sequence of other sections in the equipment and are they reporting back with good signals, I2C serial signals come to mind.

Dave

[samjmann]

Thanks everyone for such informative replies. The unit is a Technics RS-TR474M1. (NOT the M2 which has a different micro).

All the supplies are present. This includes a mains detect 5v barely smoothed DC. This is used for standby switching of the micro. The waveform on this pin is almost identical to the sketch in the manual. ThIs is pin18 POF.

The 3leg crystals voltages are both at 4.9v, which is higher than the 2.1/2.3 on the manual.

Vref: 5.7v pin 60. correct.
Vdd: 5.8v pin57. correct.

Going back to the crystal. The output impedence of the cct seems very high. Measuring with a DMM (10megohm) putting my fingers close to the test leads probe actually drops the measured voltage. This of course could be down to capacitance coupling.

I do have a short wave radio, maybe the osc frequency is way off.

There is no battery backup cap/battery in this design.

I'm getting a new xtal, for all they cost its worth a shot.

To give a little background, the deck had not not been used for years, it was then switched on and went dead. Initially there was no 5v to the micro. This was due to the 6.8 ohm safety going open in the 5.7v regulator.This was replaced, all that happened is that the standby led comes on now from a DC switched supply on the display pcb. This is not a output from the micro.

I suppose the micro could have pulled excess current and popped the safety resistor. But the replacement is OK and not running warm.

If it does turn out to be the IC, I'm sure Panasonic have long since binned all stocks. A scrap unit would be a possibility.

A close up of the Ic is attached.

Thanks again everyone. SJM.

[Maarten]

That would be a Mitsubishi (Renesas) M38123M4-101FP mask programmed microcontroller. 101FP designates the mask version. 3812 designates the product group, 3 probably the amount or RAM or other internal configuration.

I could not immediately find a datasheet for the 3812 group, but for the at least somewhat similar 3818 group. That datasheet mentions that the clock only starts after the controller comes out of reset. I suspect the same would be the case for this controller.

[samjmann]

Further to the last post, I did check the reset pin with a DSO. It was a clean 5v, followed by a sharp 0v excursion held at this then a rise back to 5v. I'm assuming that this is correct.

[samjmann]

Thanks Marten for looking into this for me. Maybe the micro is in Sleep Mode due to the POF waveform being slightly wrong.

The deck DID run. keyscan, and fluorescent display all OK. Suddenly it all went dead and then no 5v rail. From past experience I've changed Ic"s such as this and it's had no effect.

I'll try to get a snapshot of the reset pulse. I've still got the L plates on driving a modern LCD 'scope.

Once again, thanks. SJM.

[SiriusHardware]

I haven't looked, but the majority of microprocessor reset inputs are active-low (normally high, low to reset, then back high for normal running). Normally an active-low reset pin will have a name something like NRST or RESET or RST, the 'N' in the first example and the overscore on the second and third indicating an active low reset signal.

The discrepancy between the voltages you should have on the crystal in / out pins and the voltages you actually have does suggest a problem with the crystal or the oscillator section on the uP. In fact, to have both pins at exactly the same voltage would suggest that they could be directly connected together, ie, a short between the pins or short in the crystal, despite the very high impedance you have observed down to ground. Have you checked that?

Obviously you should try the new crystal first, but if that doesn't work you might just be able to rescue it by putting the crystal in an external oscillator circuit and feeding the output of that into the OSC-in input.

[samjmann]

Thanks Sirus. The thing that is most wrong is the osc cct itself. I'll get the unit back together tomorrow then post back with findings. The tape decks are really quite good. Pressure rollers and heads both with negligible wear, would be good to see it working again.

Can see the draw back to vinyl, but cassettes never quite wooed me the same way. Still we are all different!! SJM.

[trsomian]

Can you temporarily inject a 6MHz clock into the input pin and see whether it breaks into life?

[woodchips]

3 leg crystal? Sure it isn't a ceramic resonator?

[cmjones01]

Originally Posted by woodchips

3 leg crystal? Sure it isn't a ceramic resonator?

That's a good point. I've found ceramic resonators to be a good deal less reliable than quartz crystals.

Chris

[GMB]

But you do get 3-legged crystals where one is just the case.

I once encountered a batch of crystals where it was sometimes necessary to "jump start" the new crystal (so maybe also one left for a long time). I would briefly touch another crystal across it and they would burst into life. Once going they seemed to behave normally thereafter.

[samjmann]

I've just checked the device. The centre leg is grounded. Each of the two others read around 3.8meg to ground. Between the two outer legs there is a 29.8meg reading. It's actually drawn as a resonator, see my rough sketch.

Unfortunately I haven't access to 6.0mhz generator, so I'm stuck there
I will try what sirus has suggested with a SW radio tuned around 6mhz.

Once again thanks for all the help, much appreciated. SJM.

[Jon_G4MDC]

RE Post #16. It's not unknown for crystals to get "sleeping sickness".

Over time impurities settle on the quartz and dampen the Q to a point where the oscillator can no longer start them. The jump start trick using a second crystal of the same frequency or a signal generator does work but the trouble can return.

[G6ONEDave]

It could be the crystal being lazy. Not sure if I have the correct version of the service manual, there's no M1 or M2 mentioned so probably the first issue. Anyway my previous comment about the I2C system appears to be on pins 20 and 21 of the micro. Look for shorts or fixed 5v, that could lock up the micro. Also check pins 61 and 62 which is where the push button controls connect to the micro, I would expect a highish resistance across them until a button is pressed. If a button has become stuck on or even shorted at power up then the micro could hang as a result. This all depends on the internal configuration of the micro that may or may not be upset by problems of this sort.

Dave

[samjmann]

Thanks Dave.

The A/D encoder line for the various switches is high. When power is pressed it's 0v for on as in the manual. The I2c lines are both high. But I'm not sure if this a forced high due to internal breakdown or a 'soft 5v' fro a pull up resistor.

According to the service manual, reset line low activates the micro. I did have a waveform much as the manual.

The odd thing was it did work, display all OK etc, then dead. That's when the excess current went via the 5v regulator.

If anyone wants the manual, Hi-Fi engine has it. This is the version 1. It has a coloured manual on the site. It's complete but awkward to follow due to been scanned in.

This is the suspect resonator in the picture

Regards, SJM.