hi everyone i need 16 bit parallel eeprom. Anyone know where i can find this IC? if there's no 16 bit parallel eeprom, can i cascade 2 8bit parallel eeprom? i need this eeprom to make sinusoidal signal, i'll combine this eeprom with 16bit dac. any other suggestion?
Hi. First, why do you need an EEPROM (electrically eraseable) when an EPROM (electrically programmable/UV eraseable) will do -- after all, the data for a sine wave isn't going to change or need to be reprogrammed on the fly.
If you need a 16-bit wide data space with 2^16 (64K) positions, get two
27C512 EPROMs (64K X 8) and program one with the high data bytes, one with the low data bytes. Of course, make sure to deal with the control lines, too.
Right. As a practical matter, 16 bits is overkill for sine wave generation.
It has a number of practical disadvantages, too. For a 16-bit waveform, you'll have to run the counter clock at 65536 times the sine wave output frequency. Even a faster EPROM with 100ns access will give you a maximum output frequency of only 152 Hz. That also assumes you don't have any issues with the speed of the DAC, which may slow you down more..
You'll probably find that a real world "16-bit" DAC will only give you
13 or 14 bits of resolution. Even if the output is monotonic (i.e. the output is guaranteed to increase when the digital input increases), internals in the DAC will hurt you here.
Unless you're looking at testing the 16-bit DAC (this might be educational, if you've got a scope with delay), it might be better to go with the above advice, and use a 12-bit DAC with some filtering.
wow, thanx for reply, its very helpfull for me and give me some consideration about my project. I'm sorry about my english before, its not grammatical and very bad i think. OK this my real problem. I need to make 9 phase sinusoidal wave generator. This wave generator can also be operated on 5 phase. Frequency for this system is on the range about 0 Hz till 150 Hz. I have two plans to solve this problem First i'll use VCO to generate clock for one counter which serve address data for 9 eeprom. So each eeprom will contain data for each phase of sinus. Data from eeprom is then converted to analog with dac. The problem in this system is i can't switch my 9 phase system to 5 phase system, because the data in eeprom is fixed. Second, i'll use microcontroller as clock generator for counters, there are nine counters to serve nine address eeprom. So each eeprom will contain similar data that is one sinusoidal data.The phase control is arranged by microcontroller. The reason why i use 16-bit system is for precision. Any comment for that?
If English is your second language, you're doing far better than I could.
If I had to do something like this, I would try like heck to get the spec lowered to 15 bits (c'mon, that's only 150 _microvolts_ per step for a 5V p.p. signal -- who needs even that much precision?), and then use the 16th bit on each of the 27C512s to switch over between the
9-phase and 5-phase data array. That way, you would have high byte and low 7 bits of one output (either 9-phase or 5-phase, depending on the high bit) on two EPROMSs. You could then have your standard DAC output. Of course, four of your EPROMSs would have all zeroes programmed in on the 5-phase output.
Here's what I mean.
0x0000 - 0x7FFF = 9 phase data for one output
0x8000 - 0xFFFF = 5 phase data for one output
A15 on the EPROM is used to switch between 9-phase and 5-phase.
This means that you're going to use a VCO (1 IC plus passive components), a 15-bit binary up-counter (2 ICs), 18 EPROMs and 9 DACs. You could use a switch, a logic input, or a microcontroller output port pin for the control of the 9- or 5-phase bit on the EPROMs. I wouldn't add another eight counters, that will only make your 30 IC Acres O' Chips bigger if this is a real-world project. But there are other less wasteful ways to do this, particularly if you've got a microcontroller or are willing to accept somewhat lower resolution on your DAC.
16 bit DAC is not common component here, i couldn't find this IC. TMS27C512 EPROM is so expensive here, i guess this not common component too. Winbond EPROM cheaper.I don't understand why these two ICs' prices are awfully significant. Is texas instrument quality better than winbond? Do you really think that 8-bit system is enough for sinus wave generator? about problem that i have to run the counter clock at 65536 times the sine wave output frequency, i think its depend on how much the data or sampling of sinus. If i just use 20000 data, its mean that i only have to run the counter 20000 right? The precision that i need is the value of quantitation of sinus. 16 bit is better than 8 bit right, especially when i want to analize an extreme condition like near zero condition of sinus or on peak value of sinus. Do you understand what i mean? (I hope my write isn't wrong)
;edit version :) thanx once again for reply. i've got another problem
16 bit DAC is not a common component here, i couldn't find this IC. On the other hand, TMS27C512 EPROM is so expensive around here, i guess this not a common component either. Winbond EPROM is cheaper. I don't understand why these two ICs' prices are awfully significant. Is texas instrument's quality better than winbond's? Do you really think that 8-bit system is enough for sinus wave generator? about problem that i have to run the counter clock at 65536 times the sine wave output frequency, i think it depends on how much the data or sampling of sinus. If i just use 20000 data, it means that i only have to run the counter 20000 times right?I need a precise quantification of sinus. I suppose 16 bit is better than 8 bit, especially especially when i want to analyze an extreme condition like near zero condition of sinus or on peak value of sinus. btw, thanx for idea to separate eprom, its brilliant idea. :)
Hi. Reality tends to intervene in these things. 16-bit DACs are _very_ expensive. 12-bit DACs are more reasonable in cost.
You really haven't described the end use of this 9-phase/5-phase sine wave generator, so we can't really give you good advice on the resolution you need.
IN order to utilize the investment you make in a DAC, the count should be equivalent to the resolution. If you've got a 16-bit DAC, you should use a count of 2^16 = 64K. If you've got a 12-bit DAC, 2^12 =
4096 or 4K. For an 8-bit DAC, 2^8 or 256. Obviously, the greater the count, the greater the practical limitation on output frequency.
Since you're mentioning Winbond, I would hazard a guess that it's an audio application. 12 bits sounds like more than enough for anything audio. Remember that your distortion coming from the DAC is going to be based in large part on the resolution. 12 bits means a count of
4096. This means that you'll have a count step of just a little over 1 millivolt for a 5V p.p. signal. That's really trivial for audio.
2764 and 27C64 100ns EPROMs are also dirt-cheap, and will provide you with an 8K X 8 data space. If you go with those and a 12-bit DAC, you can use the same bank switching scheme (top bit A12 is the switch between 9-phase and 5-phase) to reduce the cost and shrink your "Acres O' Chips" by a bit like this (view in fixed font or M$ Notepad):
Since you've got an odd number of DACs, you'll be left with a half-used EPROM, but this will get the number of EPROMs down from 18 to 14, so your total number of ICs on board should be reduced to 26 from 30.
If nanosecond-level glitches from the counter are a problem, you'll also have to put together a strobe which follows the clock, to grab data coming from the counter after the count is set. But you also might just want to use a little output filtering on the DAC to knock this down.
Again, this is a brute-force method which is not elegant at all. There are better ways to do this, particularly if you have a microcontroller available. But in order to get help, you have to provide more information. The more the better. If a couple of minutes with a piece of paper writing down your project requirements and another couple of minutes typing it in will save you a dozen hours on the bench and reduce your project cost, it might be worth it.
Yes, but not on the 16-bits issue - you don't need 9 eproms, or even 5.
What you need is one eprom, nine latches, and a little clever programming.
You would run nine concurrent "processes", which are basically ring buffer counters, offset from each other by 1/9 full-scale, so to speak. For the five-phase, you just use the first 5 latches, and set your first 5 processes offset 1/5 full-scale from each other, and set the other 4 idle.
Oh my I'm sorry for not telling you all my complete problem, i don't expect that my problem really complex. Ok these are my problem. I'm doing final project now, its about multiphase system. I want to analyze ripple generated from power inverter. Inverter i must build is 9 phase inverter and the control method i used is PWM method. So i need to generate 9 phase sinusoidal signal and one triangular signal. Then this two signal are compared and the result is used to switch the IGBT power switch. That's all.