How about linearity at higher Iq ?
PS:
2SA1381/2SC3503 are easily available at fair prices, 300V versus 200V for the A1380/C3502.
The 300MHz Sanyo 2SA1540/2SC3955 Pavel likes so much can both still be had in GB, some suppliers offer at even lower rates than Nikko's.
Hi Jacco,
your comment/question
Do you want/need to elaborate for us?
your comment/question
implies that transistor characteristics, in addition to the list from Pooge, do have an influence on the GBW product and the Slew rate.
Do you want/need to elaborate for us?
Re: schematic
Thanks! It'll be good to look at these and get a sense of how it's evolved.
loek said:
Thanks! It'll be good to look at these and get a sense of how it's evolved.
AndrewT said:
Actually, you may have to increase the lag cap to make up for the lower Cob of transistors used, to maintain the same GBP. It is the parallel combination of the lag cap and Cob that is the factor. This isn't a bad thing, since the added cap is more linear than the transitor Cob. Therefore, lowering Cob and raising the lag cap to get the same overall capacitance is good because it reduces modulation in the VAS.
AndrewT said:
Andrew, don't know where you are getting these numbers. For an output current of 5.4A and hfe of 60 in the output transistors, the drive current into the bases of the two output transistors would total 5.4/60= 90ma. Don't know where your 400 to 500ma came from.
Also, I am wondering if your analysis is based on the characteristics of a voltage gain stage rather than a current gain stage that it is. I'm not sure that the drivers are limited to the quiesent current. I think Leach just showed in those diagrams that they never shut off at zero crossing. These drivers will easily handle 1A or more, providing plenty of room at the opposite end.
Hi Pooge,
I'm glad to see at least someone is prepared to read my hypothesis and be ready to argue the point being made.
The 400mA to 500mA figure I have used
PS. do you type the author in your quoted extract or is that generated by the Forum Software?
I'm glad to see at least someone is prepared to read my hypothesis and be ready to argue the point being made.
The 400mA to 500mA figure I have used
refer to LOAD current not base current. At this level of output current the amp is delivering just 800mW of sinusoid power into an 8ohm load. At the Iq level of 140mA the output stage goes into ClassAB at about 280mA (or slightly less) and drives 310mW of sinusoid power into an 8ohm load.
PS. do you type the author in your quoted extract or is that generated by the Forum Software?
Not feeling really competent in this subject to argue. Just guessing. I think class A refers to the fact that the opposite sex driver never cuts off, unlike the output transisters. I also don't think the total amount of current in the drivers is limited to the bias current. I think the predrivers can operate to direct more current flow in the drivers. In other words, I think the normal sex driver can keep increasing its load current while the opposite sex driver never shuts off, i.e., never goes below the bias current.
AndrewT said:
I didn't know what you meant at first. I assume you mean the "Originally posted by AndrewT" part of the quote? This is automatically inserted by forum software when I check the quote button at the bottom of your post.
Hi,
thanks.
For ClassA the single ended stage runs at Iq and Iq is modulated up or down to reach almost zero and almost double Iq when driven to the output voltage limit. If driven any harder severe distortion sets in due to clippng of the output signal.
A similar mode of operation follows for a Push Pull ClassA stage.
The Iq flows from upper to lower devices. When an output is required the upper device sources extra current and the lower device sinks less current on the +ve half cycle. This balanced operation is fundamental to Push Pull ClassA.
The difference between upper and lower current is fed to the load. At the limit the upper device sources almost twice Iq and the lower device sinks almost zero Iq. At these currents the load is taking almost twice Iq. This is the limit of ClassA operation.
If the load needs extra current the Push Pull stage can source extra current to the load and the lower device does no more to help this supply to the load. The -ve half signal follows the same process. The lower device sinks almost twice Iq and the upper device sources almost zero current forcing the load to supply the reverse current to the lower device. This corollary applies to all the following argument when -ve portions of the signal are being amplified.
Now apply this logic to the drivers in the dual EF stage.
If the driver Iq is modulated by nearly the Iq limit then the upper device sources almost twice Iq and the lower device almost sinks zero current. This is true ClassA.
If the lower device now takes no further part in sinking current from the upper half then this mode mimics ClassAB. Lets go a bit further. The Leach diagram shows the upper device sourcing more current to the base of the output stage with the lower device staying at constant current. The lower device is NOT passing ZERO current but it is NOT helping the upper device either. This mode exactly mimics ClassAB. That is my contention.
Whether the lower device sinks zero current or simply stops modulating it's output , makes no difference to the fact that the upper device does all the work when current demand exceeds twice Iq. This is ClassAB. Leach may say that the device does not switch off, but that is not the definition of ClassA.
Progressive and balanced control of source and sink currents is demanded by a Push Pull stage for it to remain in ClassA mode of operation.
Operation of a single source (or a single sink) in a push pull stage is classAB operation. Everyone seems to agree this in the output stage. Why should it be different in the driver stage?
thanks.
I think you are almost right but falling into the same trap as Leach.
For ClassA the single ended stage runs at Iq and Iq is modulated up or down to reach almost zero and almost double Iq when driven to the output voltage limit. If driven any harder severe distortion sets in due to clippng of the output signal.
A similar mode of operation follows for a Push Pull ClassA stage.
The Iq flows from upper to lower devices. When an output is required the upper device sources extra current and the lower device sinks less current on the +ve half cycle. This balanced operation is fundamental to Push Pull ClassA.
The difference between upper and lower current is fed to the load. At the limit the upper device sources almost twice Iq and the lower device sinks almost zero Iq. At these currents the load is taking almost twice Iq. This is the limit of ClassA operation.
If the load needs extra current the Push Pull stage can source extra current to the load and the lower device does no more to help this supply to the load. The -ve half signal follows the same process. The lower device sinks almost twice Iq and the upper device sources almost zero current forcing the load to supply the reverse current to the lower device. This corollary applies to all the following argument when -ve portions of the signal are being amplified.
Now apply this logic to the drivers in the dual EF stage.
If the driver Iq is modulated by nearly the Iq limit then the upper device sources almost twice Iq and the lower device almost sinks zero current. This is true ClassA.
If the lower device now takes no further part in sinking current from the upper half then this mode mimics ClassAB. Lets go a bit further. The Leach diagram shows the upper device sourcing more current to the base of the output stage with the lower device staying at constant current. The lower device is NOT passing ZERO current but it is NOT helping the upper device either. This mode exactly mimics ClassAB. That is my contention.
Whether the lower device sinks zero current or simply stops modulating it's output , makes no difference to the fact that the upper device does all the work when current demand exceeds twice Iq. This is ClassAB. Leach may say that the device does not switch off, but that is not the definition of ClassA.
Progressive and balanced control of source and sink currents is demanded by a Push Pull stage for it to remain in ClassA mode of operation.
Operation of a single source (or a single sink) in a push pull stage is classAB operation. Everyone seems to agree this in the output stage. Why should it be different in the driver stage?
There are different definitions of Class A.
See, for example, Elliot's class A page:
http://sound.westhost.com/class-a.htm
where he defines class A as conducting for 360 degrees, or never reaching cutoff or saturation.
While your definition is one of the definitions, I think it is more common in voltage gain stages rather than current gain stages. In the amp output, the voltage gain is around unity. Also, remember that current from the predriver is also available.
See, for example, Elliot's class A page:
http://sound.westhost.com/class-a.htm
where he defines class A as conducting for 360 degrees, or never reaching cutoff or saturation.
While your definition is one of the definitions, I think it is more common in voltage gain stages rather than current gain stages. In the amp output, the voltage gain is around unity. Also, remember that current from the predriver is also available.
pooge said:
this is what i understand to be the way those drivers operate, jim bongiorno also explained it like this for his GAS amps..
Hi,
this is one area where ESP got it completely wrong. He had to rewrite his ClassA section to correct the faults and misunderstandings of how a Push Pull ClassA stage worked.
I believe he still does not UNDERSTAND this part of the subject and for this reason I could not recommend ESP as a source for ClassA information.
this is one area where ESP got it completely wrong. He had to rewrite his ClassA section to correct the faults and misunderstandings of how a Push Pull ClassA stage worked.
I believe he still does not UNDERSTAND this part of the subject and for this reason I could not recommend ESP as a source for ClassA information.
Hi,
Class A is all about current NOT voltage.
Change from high ohm speakers to low ohm speakers and an output stage that was pure ClassA for all output powers becomes ClassAB when the stage output current exceeds the ClassA current.
Class A is all about current NOT voltage.
Change from high ohm speakers to low ohm speakers and an output stage that was pure ClassA for all output powers becomes ClassAB when the stage output current exceeds the ClassA current.
Hi,
the input resistor to ground r17 is usually set to 22k to match the NFB resistors.
I have built two versions and both have a 20 to 25mV output offset.
If I change r17 to about 33k the offset moves to almost 0mV and seems to hold quite steady over a range of temperatures.
Why is it necessary to UNmatch the input and NFB resistors to zero the output offset?
Is there a component I have used that is causing this slightly high offset? Where should I look.
The 33k for r17 better suits my input filter ( I would prefer 51k) and I'm likely to keep it this way, but I would like to understand what is causing the offset.
the input resistor to ground r17 is usually set to 22k to match the NFB resistors.
I have built two versions and both have a 20 to 25mV output offset.
If I change r17 to about 33k the offset moves to almost 0mV and seems to hold quite steady over a range of temperatures.
Why is it necessary to UNmatch the input and NFB resistors to zero the output offset?
Is there a component I have used that is causing this slightly high offset? Where should I look.
The 33k for r17 better suits my input filter ( I would prefer 51k) and I'm likely to keep it this way, but I would like to understand what is causing the offset.
R17 and R11 of Jens' boards match R1 and R2 of Leach's, according to my notes.
It is R17 and R11 that should ADD to 22k to match feedback network R12 + R13 of 22k.
This is what Leach specified.
Also, you should measure this with the input open. If you short it, it will upset the balance.
I'm not sure why this specification is so important, because when you connect to the output impedance of the preamp or the like, you are putting in a parallel impedance of 100 ohms or so, which, in combination with R17, reduces the value below 100 ohms. I fell into this trap of measuring when I added an input transformer, which output impedance threw off the DC balance.
It is R17 and R11 that should ADD to 22k to match feedback network R12 + R13 of 22k.
This is what Leach specified.
Also, you should measure this with the input open. If you short it, it will upset the balance.
I'm not sure why this specification is so important, because when you connect to the output impedance of the preamp or the like, you are putting in a parallel impedance of 100 ohms or so, which, in combination with R17, reduces the value below 100 ohms. I fell into this trap of measuring when I added an input transformer, which output impedance threw off the DC balance.
Andrew,
I was going to suggest that you measure the offset with the input shorted to ground through a cap.... this will ensure that you don't have different base resistors to ground on the two sides of the diff amp.
\Jens
I was going to suggest that you measure the offset with the input shorted to ground through a cap.... this will ensure that you don't have different base resistors to ground on the two sides of the diff amp.
\Jens
If this were implemented then r17 would be less than 22k. That in turn would make the output offset worse not less.
Quite why Leach arranges his input components this way is a mystery but he must have had a reason/s.
The Leach arrangement , which many others also adopt, deliberately unbalances the input conditions on the two sides of the LTP.
But I come back to the "understanding" of why the need to unmatch the input impedances is necessary to zero the output offset.
Hi,
Similarly DC couple the NFB and AC couple the input does not balance out easily.
Either AC couple or DC couple . Do not mix or you may have wierd DC output offsets at various times from cold startup to quiescent running to gentle background to full on party night (annoy the neighbours) temperatures
Warning !!! DC couple and then inadvertently add a DC blocked output from another pre-amp and at a stroke you have a mixed AC/DC coupled amplifier. Take care, or install two input RCAs, one DC input and the other AC input which Jens allowed for on his PCB.
you are into the realms of partial DC coupling i.e. DC couple the non-inverting input but AC couple the NFB. This is really difficult to balance out and retain sensible impedances around the LTP stage. Don't!!
Similarly DC couple the NFB and AC couple the input does not balance out easily.
Either AC couple or DC couple . Do not mix or you may have wierd DC output offsets at various times from cold startup to quiescent running to gentle background to full on party night (annoy the neighbours) temperatures
Warning !!! DC couple and then inadvertently add a DC blocked output from another pre-amp and at a stroke you have a mixed AC/DC coupled amplifier. Take care, or install two input RCAs, one DC input and the other AC input which Jens allowed for on his PCB.
- Status
- Not open for further replies.