Ive been curious a long while now, after working with the 723 on several projects, I must say Im highly impressed with the 723 in terms of price vs. performance, Ive used it several times in HV applications particularly tube based preamps.
There have been a handful of companies that have used the 723, but I'm curious is to why Audio Research in particular didn't implement it in any of their designs, they chose to use the LM329 as a VR in just about every preamp and pre section of Their amplifiers, and its not near as good as the VR in the 723, even with its low grade internal op-amp, if its Vcc is buffered with a BJT it can achieve a 2uV noise level. It doesnt make any sense to Me. Someone help Me out here.
There have been a handful of companies that have used the 723, but I'm curious is to why Audio Research in particular didn't implement it in any of their designs, they chose to use the LM329 as a VR in just about every preamp and pre section of Their amplifiers, and its not near as good as the VR in the 723, even with its low grade internal op-amp, if its Vcc is buffered with a BJT it can achieve a 2uV noise level. It doesnt make any sense to Me. Someone help Me out here.
I don't have an answer to your question but.
So far, I have made a pair of HV regulators from Audio Research LS25MKII and I am very satisfied with them and they are much better than the classic solutions and other regulators for HV. The same regulator can be improved by replacing the LM329, op amp and adding two new current sources. You can see the difference in the pictures, measured in my DAC while all digital is working, 260V, 100mA.
See this and there will be more soon; Walt Jung reference
Igor
Attachments
Nice work!! Using a current source for those 329's and the Op-amps would have been a huge improvement.
The 723 was a good chip for its time, but it was relatively quickly displaced by other circuits. For simplicity, easiness in POL applications, 3-terminal regs took over, and for high-end applications, LM199 and the like (I think the 329 is a downgraded 199 without thermal control) were preferred, because of the performance of the sub-surface zener (the technology didn't exist when the 723 was introduced).
The 723 remained popular for industrial supplies exceeding the capabilities of the 3 pin regs, but was ousted when SMPS became popular.
It is possible to build a very good regulator based on the 723; it won't be as efficient or low-noise as modern alternatives.
The guaranteed noise spec of a 329 is vastly superior compared to the 723, but the raw figure can be improved with a good circuit.
Note that this remark is valid for any circuit: @Trileru variations on the denoiser theme are ~200 times better than a raw 329, which itself is 4 times better than a raw 723, yet the 317 is basically worst than both
The 723 remained popular for industrial supplies exceeding the capabilities of the 3 pin regs, but was ousted when SMPS became popular.
It is possible to build a very good regulator based on the 723; it won't be as efficient or low-noise as modern alternatives.
The guaranteed noise spec of a 329 is vastly superior compared to the 723, but the raw figure can be improved with a good circuit.
Note that this remark is valid for any circuit: @Trileru variations on the denoiser theme are ~200 times better than a raw 329, which itself is 4 times better than a raw 723, yet the 317 is basically worst than both
From the datasheet the best on the 329 is 7uV, the 723 is 2uV that is with no filtering on the REF, and a buffer on the Vcc, if one where to filter the REF as well, who knows the figure, the cost vs. performance is overwhelming!!! The 723 is the clear winner!!
Roger Modjesky if alive , would confirm my analysis as well.
Roger Modjesky if alive , would confirm my analysis as well.
LM723 is an excellent regulator chip in this time that is still being used in new designs and this dinosaur still is in production (since 1969). That says something.
Personally I have had best results with regards to low noise and DIY PSU's with LM723 but I regret it is not LDO. The noise numbers are way better than most modern 3 pin LDO regulators. Opinions are opinions but facts are facts. As usual: many won't like it that granddad finishes the 2022 Tour de France as second or even first.
Component count is higher than we are used to today, an external pass transistor is also unusual today. It surpasses most 3 pin regulators easily even when the layout is off. With a handful of parts one can DIY a relatively simple ultra low noise linear PSU that operates stable without nasty sideffects for decades. A hard to beat and usable PSU that is. Given it is not LDO one should always calculate losses in heat. Low power applications seem most appropriate.
Let's put it this way: chances are high that a DIYer can build a stable well performing ultra low noise PSU with this IC even without a ton of measuring equipment or a lab. It won't have output peaks at power on, it will not oscillate, it will simply give clean power.
Personally I have had best results with regards to low noise and DIY PSU's with LM723 but I regret it is not LDO. The noise numbers are way better than most modern 3 pin LDO regulators. Opinions are opinions but facts are facts. As usual: many won't like it that granddad finishes the 2022 Tour de France as second or even first.
Component count is higher than we are used to today, an external pass transistor is also unusual today. It surpasses most 3 pin regulators easily even when the layout is off. With a handful of parts one can DIY a relatively simple ultra low noise linear PSU that operates stable without nasty sideffects for decades. A hard to beat and usable PSU that is. Given it is not LDO one should always calculate losses in heat. Low power applications seem most appropriate.
Let's put it this way: chances are high that a DIYer can build a stable well performing ultra low noise PSU with this IC even without a ton of measuring equipment or a lab. It won't have output peaks at power on, it will not oscillate, it will simply give clean power.
Last edited:
since the 723's internal output stage has an open collector npn series pass transistor, i think that with some cleverness and a few additional parts, with extra care to design the frequency compensation, a low vcesat pnp or even a pmos transistor could be used as an external pass device for an ldo. Perhaps not worth it given that there are plenty of purpose–designed ldo controller ics out there, but maybe a fun side project which might end up usable
Hi, I used it in a few designs and one of those was measured by a member here. It had lowest noise of all tested PSUs then and it had no side effects. I did not see a need for trying to improve as it already had best of class numbers then.
For reasons I am probably too stupid to understand SMPS are now the de facto standard for low power applications at diyaudio.com. This dinosaur IC in a simple design betters almost any consumer grade SMPS big time. Absolutely no realistic comparison as the difference in ripple/noise is quite absurd. Think of factors in the hundreds to even thousands 🙂
As the people that think they're important try to convince that SMPS is the way to go I pinch myself in the arm regularly and I compare LM723 based PSU's with various SMPS as sold with DACs, audio players etc. Outcome is always the same till today. I prefer to use current technology parts but LM723 is hard to overlook.
For reasons I am probably too stupid to understand SMPS are now the de facto standard for low power applications at diyaudio.com. This dinosaur IC in a simple design betters almost any consumer grade SMPS big time. Absolutely no realistic comparison as the difference in ripple/noise is quite absurd. Think of factors in the hundreds to even thousands 🙂
As the people that think they're important try to convince that SMPS is the way to go I pinch myself in the arm regularly and I compare LM723 based PSU's with various SMPS as sold with DACs, audio players etc. Outcome is always the same till today. I prefer to use current technology parts but LM723 is hard to overlook.
Last edited:
It depends on the source:
This one states 86µV typical, unfiltered for the 723.
This one gives 7µV for the 329:
https://www.analog.com/media/en/technical-documentation/data-sheets/129329fd.pdf
In reality a simple low power LM723 PSU design, so not just the IC itself, can be below 20 µV of noise (often lower but it depends on brand/pass transistor and load current). No trickery, just datasheet schematic of course with Cref.
Maybe comparison of LM723 with LM329 is odd as LM329 is only a Zener reference while LM723 has a band gap reference built in together with surrounding components to be a voltage regulator IC. For comparison a complete PSU design with LM329 and one with LM723 should be compared I think.
When wanting to know what it is capable of it is recommended to use good IC sockets as differences between brands are an issue. To our luck industrial PSU's with it generally have IC sockets.
Maybe comparison of LM723 with LM329 is odd as LM329 is only a Zener reference while LM723 has a band gap reference built in together with surrounding components to be a voltage regulator IC. For comparison a complete PSU design with LM329 and one with LM723 should be compared I think.
When wanting to know what it is capable of it is recommended to use good IC sockets as differences between brands are an issue. To our luck industrial PSU's with it generally have IC sockets.
Last edited:
AFAIK the National and/or ST produce(d) versions that do have a bandgap reference. That was one of the innovations at that time. It is too long ago but I think it were the CH and J types. NTE923 (equivalent type) also has it. I have quite some versions of the chip NIB/NOS of many years just because they're way above the general 3 pin regulators, take no space and they're intriguing parts.
I just checked and these are still available (and both "active" and "bestseller" 😀) but at 13 Euro a piece. The commonly used CN version with quite good temperature-compensated Zener diode is 1 Euro. For 1 Euro their performance is not shabby either.
I just checked and these are still available (and both "active" and "bestseller" 😀) but at 13 Euro a piece. The commonly used CN version with quite good temperature-compensated Zener diode is 1 Euro. For 1 Euro their performance is not shabby either.
Last edited:
https://electronicprojectsforfun.wo...esign-practices-using-the-lm723-ic-regulator/
Here is great article on how to get the most out of the 723.
I dug a little deeper, and found that the reality is even weirder: those "irregular" chips not only used a bandgap ref, but they also kept the zener -albeit a lower voltage one-.
The 5.9V zener has a ~zero tempco, and so has the 1.25V bandgap, resulting in an identical final voltage and ~zero tempco.
This is strange, as it is normally easier to manufacture a zener having a well-defined positive tempco than aiming for zero, which is ill-defined.
All the traditional, high accuracy zeners used a higher voltage than 5.9V, combined with one or two forward-biased junctions for that reason.
Some details here for example:
https://www.eevblog.com/forum/projects/lm723-die-pictures/100/
One really needs to read the original datasheet for the 723 to get a better grasp on it. Its a super primer on regulators as well.
https://electronicprojectsforfun.files.wordpress.com/2018/08/ua723_application_note.pdf
https://electronicprojectsforfun.files.wordpress.com/2018/08/ua723_application_note.pdf
What baffles Me is why they didnt update the chip with a better internal error amp, One that doesnt need an external freq. comp. capacitor, but that also changes the price of the chip considerably I would first guess.
For understanding LM723 one must be prepared to do some reading and one will encounter some surprises and datasheet errors. Literally all has been said repeatedly the past decades. It would be a nice practical challenge to design a 2022 linear PSU with it that can compete with todays regulators. I know it can in some aspects and I know how the reactions were when I did. Truth is that I still get requests for that PSU as it eats SMPS for breakfast.
When (re)reading up on LM723 I learnt that a 250 mA 4.5 ...50V version with lower dropout voltage exists. Too bad it has more noise.
Anyway, contrary to popular belief the chip is alive and kicking and it is being used in various equipment by many manufacturers. So the measuring equipment built the last 35 years likely has LM723 in it for lowest (flicker) noise and we use it in audio to measure circuits that either have 78xx, LM317 or the now popular extremely noisy SMPS 😀
BTW it really was time that TPS7A4700 and LT3042/3045 were invented.
When (re)reading up on LM723 I learnt that a 250 mA 4.5 ...50V version with lower dropout voltage exists. Too bad it has more noise.
Anyway, contrary to popular belief the chip is alive and kicking and it is being used in various equipment by many manufacturers. So the measuring equipment built the last 35 years likely has LM723 in it for lowest (flicker) noise and we use it in audio to measure circuits that either have 78xx, LM317 or the now popular extremely noisy SMPS 😀
BTW it really was time that TPS7A4700 and LT3042/3045 were invented.
Last edited: