Bingo.
Bear, separate power supplies for different stages in a signal path can have major benefits, especially when high currents and Class-AB output stages are involved. However, this is also done inside op amps, many circuits inside an op amp use a subregulated power supply. Also, several op amps internally generate an different supply rail for the input stage (commonly called zero input crossover distortion devices). Furthermore, op amps have been offered which would allow for a separate supply to be used for the output stage, or even a fully discrete output stage (such as the LME49810, which was just a glorified power op amp with some added features).
However, I believe that if I offered an op amp with separate power pins for all of the internal stages of the device, all of the same arguments against traditional op amps would still be used against it.
At their heart of what you and SY both said though is that discrete transistor circuitry gives the designer more flexibility/freedom (SY for creating effects, you for powering each stage separately). Congrats! We've established at least one drawback of op amps in audio circuitry: they limit the designers freedom in determining every single detail of the circuit topology.
Edit: we've really established two drawbacks. The other is that its more difficult to market your product when you use op amps.
Last edited:
Johnc124, we have been trying to make linear IC's sound virtually 'perfect' for the last 50 years. We have never been completely successful, no matter how our power supplies are arranged or what IC that we use. SY just doesn't believe in differences of this sort anymore. You and I should communicate sometime.
Wouldn't this statement apply to all audio electronics, not just ICs?
I'm always open for more communication!
The IC opamps LTC1152, OPA365, MAX4162, ADA4858 all have an on-chip charge pump to generate additional power supplies besides the ones connected to I/O pins. Interestingly, the ADA4858's charge pump output is brought out to an I/O pin, so the board designer can utilize that voltage.
_
_
Attachments
Yep, add the OPA2320 (and OPA2322), the OPA2325, and the OPA2369 to that list. Scott W might be able to add any additional ones from ADI.
Wow something useful. How about some advantages? Cheap, easy, small, differential input device temperature tracking, no pcb parasitics. Economies of scale. (The hours of r&d that go into an opamp must be orders of magnitude more than any of the discrete designs out there, where they spend more on marketing). I'm sure there are more dis/advantages. Let's see more of both.
A few with separate supplies for output exist, but in general there is no benefit to be had. Think about it in all these years no one has seen fit to even make a 5534 that uses 2x the supply current and has a heroic output stage. There is no money in this.
Let's look at Mark's schematic and see the problems with it in regard to quality audio:
1. Cmos is very 1/f noisy.
2. The input topology equally mixes 4 device's noise, or doubles the intrinsic noise, compared to a single device.
3. Who needs a charge pump for audio? Yes, I have used them in other applications, but we usually make a good power supply from the get-go.
4. Single ended output stage drive means lots of extra 2'nd harmonic than what can be gotten from a balanced drive.
5. etc, etc.
1. Cmos is very 1/f noisy.
2. The input topology equally mixes 4 device's noise, or doubles the intrinsic noise, compared to a single device.
3. Who needs a charge pump for audio? Yes, I have used them in other applications, but we usually make a good power supply from the get-go.
4. Single ended output stage drive means lots of extra 2'nd harmonic than what can be gotten from a balanced drive.
5. etc, etc.
Last edited:
Most of these have a specific application or performance need. I can't think of any part that added additional supplies/pins specifically telling the customer to provide extra supplies for higher performance. With most amplifiers now needing rail to rail inputs and outputs there is no benefit to the common technique of having highly regulated different rails for the input stage. Look around here to the discrete DIY designs, many don't get anywhere near the rails due to various popular circuit tricks. Take especially popular output stages some of which need volts of headroom.
Why hasn't anyone listed the numerous hi-end/prosumer products that are full of 8-legs?
Money? Profit?
We don't care about no steekin' profit!!
No, seriously, we're DIY, we want maximum benefit from the least $$ cost, greatest simplicity, and ease of construction. Some of us will settle for maximum benefit.
Point being I'm hear to say the more often than not, you can hear just the power supply being changed, like from a well filtered passive supply to a regulated supply, and between different regulated supply schemes.
If one couldn't what the heck would be the point of those articles in Linear Audio trying to find the best performing discrete regulator methods, and THEN having an auditioning shoot out??
Seriously, the 5534 sounds not even close in terms of "everything heard" compared to the AD767 (is that the right number? The Scott W design). Not even a horse race.
----------
Johnc1234, I sent you a PM that you never responded to...?
For the 8-legs are bad crowd I would like them to consider the following test plot. This is from a line level pre-amp where you buy the board and solder up yourself. It not only uses nasty little 8-legs, it uses Pro audio nasty little 8-legs from THAT and to add insult to injury it has a SMPS on the same PCB. Must be horrid?
Now I rather like the 32 tone test that AP offers on some models. It's not perfect but as a single test to throw up any oddities I have not seen a better one yet. For this evil horrid unit you can see that any spuria are low enough that they would be below the noise floor of the room. Nothing that could be noticed to deviate from transparent. Additional benefits include 85dB input CMRR even with an unbalanced source and full pro level drive capability. All for £7 worth of silicon. No nice story but superlative performance.
You could build something also transparent for less but not to quite this level of over the top performance.
Note I don't have one of these and its not on my shopping list as my needs are different, but I do like the THAT 1200/1646 combo for input and output duties at line level.
Now I rather like the 32 tone test that AP offers on some models. It's not perfect but as a single test to throw up any oddities I have not seen a better one yet. For this evil horrid unit you can see that any spuria are low enough that they would be below the noise floor of the room. Nothing that could be noticed to deviate from transparent. Additional benefits include 85dB input CMRR even with an unbalanced source and full pro level drive capability. All for £7 worth of silicon. No nice story but superlative performance.
You could build something also transparent for less but not to quite this level of over the top performance.
Note I don't have one of these and its not on my shopping list as my needs are different, but I do like the THAT 1200/1646 combo for input and output duties at line level.
Attachments
I'm not in the anything is automatically "bad" crowd.
I'm in they all sound different crowd (until I find some that sound indestinguishable AND have no "sonic signature" of their own).
Think I will have a look-see at the THAT 1200/1646 combo you suggested.
If applicable I will put it into the current piece of gear that has apparently become "test bed" for single 8 legged creatures... if it is equal or better of the best chip I've used to day, I'm not only happy but I'll be more than pleased to say so.
How's that?
_-_-
I'm in they all sound different crowd (until I find some that sound indestinguishable AND have no "sonic signature" of their own).
Think I will have a look-see at the THAT 1200/1646 combo you suggested.
If applicable I will put it into the current piece of gear that has apparently become "test bed" for single 8 legged creatures... if it is equal or better of the best chip I've used to day, I'm not only happy but I'll be more than pleased to say so.
How's that?
_-_-
Bear, the 5534 is missing the 2MHz carrier that most DIY designs with AD797 seem to embed.
Bear, the 5534 is missing the 2MHz carrier that most DIY designs with AD797 seem to embed.
You mean of course relatively untutored DIY'ers. The Stanford Research SR1 has 37, must be fortuitous cancellation. 🙄 If it's 2MHz you are doing something wrong, it should be 60MHz.
Last edited:
The folks at THAT are friends of mine and I would venture to guess they give little credence to much of the audiophile jabber and their business model IS the audio market. I also think they are not loath to sell into smaller markets than the big semi-houses.
Was looking at the THAT 1580 datasheet just now. Looks like it could be another option for the gain stage for digital RIAA experiments, at least if I could find someone to sell singles rather than a tube at a time.
<google> I can, but QFN-16 soldering scares me, so the 1512 at £5 in singles in DIP it is from their range.
<google> I can, but QFN-16 soldering scares me, so the 1512 at £5 in singles in DIP it is from their range.
Bear, the 5534 is missing the 2MHz carrier that most DIY designs with AD797 seem to embed.
Ha ha... but if it sounds the same ad the LM<whatever> part (for all intents and purposes) are they both oscillating?
Having never had one oscillate, what amplitude is the typical oscillation?
Or is it more like a burst parasitic?
(wow, almost real technical content is suddenly breaking out! WARNING WILL ROBINSON!!)
So, how will these devices from THAT, designed for balanced operation fare when "tipped on their side" and run single ended??
_-_-
_-_-
- Status
- Not open for further replies.