[time-nuts] Tbolt failure modes
bg at lysator.liu.se
bg at lysator.liu.se
Sat May 12 14:47:39 UTC 2012
> On 05/12/2012 03:37 PM, Mark Sims wrote:
>> It sounds like the -12V supply is used to generate the DAC low voltage
>> reference. If it is floating, the DAC output will be unstable. If it
>> is at a solid voltage, the DAC output will be stable. The standard EFC
>> range for a tbolt is -5V to +5V, but since almost all Tbolts are
>> running the osc with a positive EFC voltage, it will work with the -12V
>> line grounded.
>> I suspect that although it may appear to be working, the DAC output
>> may not be as stable as with a negative rail available since whatever is
>> generating the negative DAC reference voltage will be well below its
>> dropout voltage.
>> One thing to try is to do an oscillator autotune with the -12V line
>> grounded and at -12V. Your DAC gain will probably double (or half)
>> since the DAC output range is now 0V..5V instead of -5V to 5V. The
>> tbolt firmware knows what the expected dac voltages rails are (readable
>> via an undocumented TSIP message) are and also has settings for what
>> range to allow the EFC to swing.
> Well, all this is well, but doesn't reflect well on Björns problems I
> think, I did look at his postings in this thread and he seems to have
> valid -12V.
> What is an issue is that the big ASIC seems either powerless or kept not
> Either a 3,3V regulator went down, got shorted or something, or clock
> failed or reset line not released.
> I am a bit interested in this myself, but don't have the time to drive
> down to him this weekend.
I have checked the voltages. They are the same, measuring on the defect
unit and a working one. Have also swapped the receiver boards between
different PSUs. There is no problem running a working receiver with the
PSU of the defect one. Hence I agree with Magnus and others suggesting the
CPU (big Trimble asic) is not running.
Today I do not have access to the right instruments, but just measuring
with a DMM - there are differences at TP8, TP9 and TP10. And as mentioned
previously the CPU-chip is to cold, compared to a working unit.
I do not have time to pursue the CPU problem right now. Hope to come back
later with an update.
More information about the time-nuts