Theres a thread going about 5534 opamps from different manufacturers and someone said one measured better than another. Comparing the data sheets the specs and simplified circuit diagrams from 2 different manufacturers where practically identical. And there variations also identical, BUT some of them where huge. 2 that caught my eye where input resistance typically 300k min 30k, and CMRR typical 100db min 70db! Now if you add the not given max CMRR there may be a 40db difference from one opamp to the next. I know thes fall on a bell curve and that the chances of these 2 opamps ending up side by side are vary rare but 20 db maybe not so rare and is still quite large. A good designer is aware of this and would tend to use the worse numbers for the design to guarantee the final circuit can not be worse than this. ( or measure every device used, not cheap).
Now on to opamp rolling. Joe Roller buys a couple xx opamps to try in his poorly designed "Hi-Fi" that needs high input impedance opamps to sound right. He gets the ones that have a typical 300k input but the ones he ends up with are 500k. They "sound better", these must be better opamps. if he ended up with the ones that were 30k. They "sound worse" thes must be worse Opamps. And he tells everyone his great discovery.
These variations in almost all electronics devices seemed to be overlooked by even "competent" designers and almost totally ignored by DIYers and especially part rollers.
Now on to opamp rolling. Joe Roller buys a couple xx opamps to try in his poorly designed "Hi-Fi" that needs high input impedance opamps to sound right. He gets the ones that have a typical 300k input but the ones he ends up with are 500k. They "sound better", these must be better opamps. if he ended up with the ones that were 30k. They "sound worse" thes must be worse Opamps. And he tells everyone his great discovery.
These variations in almost all electronics devices seemed to be overlooked by even "competent" designers and almost totally ignored by DIYers and especially part rollers.
A quoted figure of 30k (or 300k) doesn't mean the opamp presents that value to the outside world in the classic sense we think of. Its a 'raw' value of the internals of the chip.
The opamp we are talking about is a differential amplifier, and if we apply for example 6 volts to each input then we would expect the output (in theory) to be zero volts and for no current to flow in or between the two inputs (infinite input impedance). In practice the real opamp is less than perfect.
If you arrange the feedback to be for a typical non inverting configuration then the input impedance will be extremely high. If it were not, then all our simple gain stages would have different absolute gain levels depending on source impedance. That of course never happens.
The opamp we are talking about is a differential amplifier, and if we apply for example 6 volts to each input then we would expect the output (in theory) to be zero volts and for no current to flow in or between the two inputs (infinite input impedance). In practice the real opamp is less than perfect.
If you arrange the feedback to be for a typical non inverting configuration then the input impedance will be extremely high. If it were not, then all our simple gain stages would have different absolute gain levels depending on source impedance. That of course never happens.
Yes, feedback is the great equalizer. The best way to control device variations, but there are still differences that might effect sound quality depending on the rest of the circuit and application. One that sticks out is open loop gain. If your using feedback to set output impedance, distortion etc than a 10db variation in open loop gain will give you a 10 db change in these parameters. So the .01% THD circuit drops to .1%
To be clear, there are variations within a given vendor's own part due to production variations, and then there are a whole pile of different masks that are called "5532", sold by a handful of semiconductor manufacturers, which may or may not share behavior.
In the old days, semiconductor companies would agree to make each other's parts using the same exact masks and basic processes in order to be able to qualify the part for certain kinds of military contracts. This is how the first wave of parts from different manufacturers that had the same basic part number came about - they were actually the same masks, done under a mutual second sourcing agreement.
Later on, as the industry matured and new companies joined in, they reverse engineered popular "gumball" parts and used the well known part number in order to imply that their device is basically the same as other similarly numbered devices from other manufacturers. This is the current state of devices like the 5534 and 5532. You have a number of devices, none of which share the same masks or process, but which generally share the same basic parameters, such as unity gain bandwidth, input noise, and output drive. If a part has "5532" slapped on it, we should expect it has the same basic input stage biasing etc., but all of the internal details might be completely different, and these details sometimes matter.
So, yes, commodity "gumball" amplifiers from different vendors do vary, but usually not due to process variations or sloppiness, but instead due to the fact that today, all devices that say "5532" do not use the same masks - they're different circuits which share the same basic characteristics.
In the old days, semiconductor companies would agree to make each other's parts using the same exact masks and basic processes in order to be able to qualify the part for certain kinds of military contracts. This is how the first wave of parts from different manufacturers that had the same basic part number came about - they were actually the same masks, done under a mutual second sourcing agreement.
Later on, as the industry matured and new companies joined in, they reverse engineered popular "gumball" parts and used the well known part number in order to imply that their device is basically the same as other similarly numbered devices from other manufacturers. This is the current state of devices like the 5534 and 5532. You have a number of devices, none of which share the same masks or process, but which generally share the same basic parameters, such as unity gain bandwidth, input noise, and output drive. If a part has "5532" slapped on it, we should expect it has the same basic input stage biasing etc., but all of the internal details might be completely different, and these details sometimes matter.
So, yes, commodity "gumball" amplifiers from different vendors do vary, but usually not due to process variations or sloppiness, but instead due to the fact that today, all devices that say "5532" do not use the same masks - they're different circuits which share the same basic characteristics.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.