The OPA627 really sings

[AndrewT]

15V in lieu of 12V increases the dissipation by 120mW instead of just 96mW.
Those extra mW increase the temperature of the whole chip

Does it mean that the chip is dissipating the same mW?.............. A surface mount non-dual OpAmp (AD825) seemed OK with a 2K and ± 15V supplies. Maybe it was at 40°C.

Hi, a normal opamp runs warm when the quasi output stage is running with just a tiny bias current (my guess=200uA).
Running the chip on high voltage rails takes the opamp to much higher internal temperatures.

Now add some 100mW or so of extra internal dissipation due to converting the output stage to single ended ClassA and the chip will run hot on low voltage rails and very hot on high voltage rails.

Q.
if the resistor loaded opamp works well with 8mA. What level of bias needs to flow if the output is loaded with a CCS? Is it half? i.e.4mA into a CCS?

 

[Ryssen]

What level of bias needs to flow if the output is loaded with a CCS?

Isn´t that depending on the load?
Or did you think of something else?

 

[AndrewT]

Isn´t that depending on the load?
Or did you think of something else?

I don't think so.
The 8mA recommendation will be load dependant, but was unspecified.
For the same unspecified conditions, I am asking whether using a CCS to pull the output stage into ClassA could operate with just half the bias current.
This should be exactly the same as any conventional single ended output stage. The least efficient are the resistor tail loading. The more efficient are the CCS tail load.
I have a feeling that if a CCS were substituted for the 1k5 to 2k0 resistor then bias and heat could be halved.

If you look back at other reports you often find CCS give superior sound quality and this may be related to the lower heating within the opamp.

 

[Ryssen]

The other benefit of using a JFET instead of a plain resistor is that the input impedance of a JFET is up in the megohms range, so the op-amp still sees a near-ideal load situation.

Same as with CCS on tubes,I´ve learned.

If you put a second JFET between the op-amp's output and a JFET configured as above, you have a "cascode" configuration. (See the schematic below.) The JFET cascode has a very interesting interaction. Q1 represents a fairly stable current bias for Q2, which means that Q2 will tend to pass a fairly stable voltage level from the op-amp's output through to Q1. Conversely, Q1 now sees a more stable voltage than it would if it were connected straight to the op-amp's output, so the current flowing through it doesn't vary by as much, which helps Q2 to maintain a steady voltage level. Thus, each JFET acts to help maintain stability in the other JFET, leading to a nearly constant current level through the cascode despite input voltage variation. In my tests, current variation for a properly-configured cascode was roughly in the +/-0.5% range. That's just a few tens of microamps of current difference for 2N5484s!

More stable current with jfets,doesn´t a CCS isolate the negative from the output better?

A notable feature of the cascode configuration is that it will have a lower current draw than that of the Q1 alone. This can be an advantage or an annoyance, depending on your application.

Lower current draw,hmm,that must depend how you set it up,(resistor?)?

This is where I coppied text from:
http://tangentsoft.net/audio/opamp-bias.html

Schema

 

[KP11520]

Hyperlink to the 2N5484 JFETs page down to bottom: http://tangentsoft.net/shop/

This is how I built it (JFET Cascode) for my situation on the attachment:

I am trying to learn (and please don't tell me) what impact changing the JFETs in similar matched pairs has on the mA. Maybe it doesn't matter, I will learn that on my own as well.

Hopefully this can help someone!