We need to make pulses from -5 to +10 volts, and brainstormed this yesterday afternoon:
- posted
2 years ago
We need to make pulses from -5 to +10 volts, and brainstormed this yesterday afternoon:
On a sunny day (Sat, 09 Oct 2021 08:05:34 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:
define 'pulses'/ speed?
10k R9 and 100nF C53 is very slow, and why then that from a flash DAC... Pulses always -5 or +10? then 2 transistors SEPP.I usually get up quite early too, noise from outside, 4 to 5 hours sleep seems to be enough for me.
The typical pulse width will be about 10 us, at about 1 KHz. It applies the bias to one side of a lithium niobate e/o modulator just before a 50 ns light pulse appears. A DC bias would be bad for the modulator.
That's a 20-bit delta-sigma ADC. The RC just removes any clock feedthru. The DAC output is rarely changed... maybe once a year. The analog switch makes the pulses.
Pulse amplitude has to be programmed to find the best extinction point of each modulator.
I sleep a lot, but often take a break around 3AM. I've heard that other people do that too.
How will the op-amp react to that switched load on it's output?
piglet
On a sunny day (Sat, 09 Oct 2021 11:45:54 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:
OK...
Last night was very late too, was flipping across TV channels and hit upon a documentary about Simon and Garfunkel, interviews see them at work in the studio with tape and making 'Bridge over troubled water' Hit me like I got a present! Took me right back to the sixties working with tape. That song revealed a gospel background. After a minute in amazement I did hit the record button! Got it!
Then it was late and I had still a lot of things to do.. became way past midnight. No, no midnight interrupts no 4 hour shifts here,, 6 hours sleep now, Sunday morning, woke up from the seven church bangs at 7,
Use some bipolar analog switch?
That is one of the potential bugs. Some opamps would go nuts, not from the load as much from the charge injection spike from the analog switch. That DG333 claims 1 pC, 10 pC, or 30 pC on various parts of the data sheet.
Gotta fix that.
I was up after 11PM (can anyone survive much past that?) editing that schematic. The Brat is doing the layout this weekend, but she took time off for the Giants (yay) vs Dodgers (boo) baseball game, so I got to own the schematic for a while.
They were going to the ballpark for the game but her SO broke his foot, so they had to watch it on their enormous OLED TV.
Incidentally, I want to buy Mo a new TV. The oleds look fabulous; do the pixels wear out?
I don't see any bug other than that 10K + 100nF time constant, which at 1000us seems rather long.
The DAC is planned to be updated maybe once a year; C53 could actually be bigger. I'm trying to persuade the customer to tune extinction once a day; his programmers think the required code would be a burden to write.
The possible bugs are:
Charge injection from the DG331 back into the OPA197. I've seen opamps go crazy for many microseconds in similar cases.
U27 pushing current into the reference through R103; two channels of that. The ADR421 is maybe not able to sink current.
On a sunny day (Sun, 10 Oct 2021 07:42:20 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:
I have a big Samsung LCD, am very happy with it. My experience with OLED is limited to the small ones, and those have severe burn in when on 24/7 for a few years, replaced one in my clock, could not even read the time anymore. I have many small OLEDs in use for instruments, the aging for something not on 24/7 is of course less, and those cost next to nothing.
One problem with OLED TVs was also that the different colors (RGB) age differently. leading to color errors over time. IIRC LG fixed this by using whie OLEDs with RGB color filters, but that is of course a hack. I see big companies like Sony and Samsung moving to other tech, like QLED:
Before you buy check some reviews!
The other thing is TV?? My TV gets it signal from an about 30$ Chinese HD satellite receiver via HDMI. Of course you can even connect an USB harddisk (I have 1 TB moves on it) or a raspberry pi (tried that too). But the 'TV' part is not used now I no longer use cable, and for that cable you needed their expensive power hungry cable box. So 2 remotes in use, one for the sat box and one for the TV, and you need both, for the TV to select a source. It all depends.... Using the TV more like a monitor.
And there is MicroLED coming, seems interesting
Yes, I did raise that question about ten hours before, didn't you see it?
piglet
Yes, I responded clarifying that charge injection is a concern. Lately, some of my posts don't seem to show up.
Ah, the DAC1220. I see the silicon is up to rev "E". Those earlier revs sure were stinkers. Limit cycles. Regions of non-monotonic behavior. And every few months, every tenth-ish chip would "rail" by going to its positive max voltage and staying there until it was power-cycled. Hilarious, especially if you're a researcher whose particle just got sprung from a trap (where it had been held for months) when one of the electrodes went nuts. TI acknowledged all of the faults and kept on plugging. I haven't used the chip in a decade, so maybe it's okay now.
We've used about 400 in the last few years and I haven't heard of any problems. Maybe they fixed it.
(signalling my day job..) Hi John I would not hesitate getting an OLED TV from either LG or Sony (the two major brands who've been releasing OLEDs for many years.) OLED & LCD TVs have generally become more reliable. LG & Sony are 2 of the higher-predicted reliability brands for TVs (with edge to Sony) - per survey researcher analyses. One of our recent surveys, my colleagues in survey research reported 98%+ of all TVs bought in the last 5 years are still operating OK. Of those which did break, #1 issue at ~ 30% was, TV wouldn't turn on. #2 at ~20% was lost picture. #3 at ~10% was distorted/poor picture quality. #4 issue at ~10% was bad pixels. I know these problems may sound general, but this is the kind of survey results I get. Cheers, Rich S.
The "1pC" charge injection claim is on the first page of the data sheet.
The "10pC" maximum charge injection is on page 3 for specific conditions.
There as graph of charge injection as a function of "source voltage" on page 5. running from zero at -15V down to -20pC at -10V before rising back to zero again at 0V and going up from there to about 30pC at about 13V where it sticks until you get up to 15V.
Somebody who actually did electronic design might try to work out what was going on and try to fudge the circuit so that there wasn't much charge injection in their particular application.
There's a device level circuit diagram of the part on page 2 of the data sheet - thinking about what those parts might be expected to do could be a useful exercise.
Bitching that the data sheet didn't spoon-feed you the data you need is less useful.
By taking over Vishay and getting them to give you the kind of data sheet you'd like?
My test people report no problems with it so far.
It's used in a 8-channel isolated thermocouple simulator.
On a sunny day (Tue, 12 Oct 2021 08:04:35 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:
Who buys a 'thermocouple simulator' ? You can get a real thermocouple for peanuts, and open and short takes a second. Heating and cooling one is not that hard either, at least for me... The Matrix???
People testing aircraft systems mostly. They want to automate testing, without people dipping thermouples into buckets of water.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.