JamJar: an HPA-1-inspired power amp

Jump to Latest
#81
........back to my sandwich. 😀
 

Attachments

  • Vegemite-Sandwiches-1-of-1.jpg
    Vegemite-Sandwiches-1-of-1.jpg
    111 KB · Views: 295
#86
Pico,

I live in Northern California about two hours from San Francisco funnily enough I lived in North Carolina a few years ago, You should give us a call when you are in the neighborhood next....sandwiches on me.

I suppose the immigration department looks dimly on transportation of Vegemite across national borders.😀

Jam
 
#87
Yeah I was also in Sacramento.
Is that anywhere near you?

Don't worry I have one huge jar and another in a tube. Like toothpaste tube. Hahahahah
 

Attachments

  • 15746549077706887672308691808476.jpg
    15746549077706887672308691808476.jpg
    966.2 KB · Views: 268
Last edited:
#92
Here are two versions. The first is one of my early attempts, while the second is based on Jam's suggestion. Neither one works.

Does it have something to do with the fact that the NFB loop goes from the VAS back to the LTP? (This is a feature of the BA1/2/3 series from which this project originated.)
 

Attachments

#94
The two attachments to post #92 appear to be ideal candidates for the divide-and-conquer method of diagnosis. You could study the left half-circuit in vitro, and completely independently, study the right half-circuit in vitro. Are both of them perfect in every respect?

When they ARE both perfect in every respect, you could connect them together through a series resistor whose value is {PARAM_1} ohms. Step the value of {PARAM_1} from 1E7 to 1E-2 ohms, simulating each time.
 
#95
Grrrrrr.... turns out to be a duff SPICE model.

While I had already isolated it to the cascode, I decided to apply Mark's divide-and-conquer model to the cascode itself. First I hooked up the transistor as a diode, which should have no effect other than 1Vbe of voltage drop. No dice; wonky results.

I then put in an actual diode in place of the transistor: everything is ducky.

So I substituted a 2SC3503 for the KSC3503 (hooked up as a cascode): it works fine.

[This is where the swear words would go if they were allowed. There would be a lot of them.]
 
#96
Try replacing the leds on beefcake 5 with a resistor and keep changing the resitance values until it works.
Once you know the right voltage then try throwing back in the correct number of leds to achieve the same voltage.

EDIT: LOOKS LIKE YOU FIGURED OUT THE ISSUE
 
#97
JeffYoung said:
Grrrrrr.... turns out to be a duff SPICE model.

So I substituted a 2SC3503 for the KSC3503 (hooked up as a cascode): it works fine.

[This is where the swear words would go if they were allowed. There would be a lot of them.]
Click to expand...

You dumb bastard.
Hahahaha

I always use the standard ltspice component models and only add models that I have personally made myself or someone I trust eg Cordell etc.
 
#100
The downside of the cascode is that I need another 5V on the rail. I have a separate FE supply, so it's doable -- but it's shunt regulated and already well capable of toasting Jam & Pico's vegemite sarnies.

I could go to a BJT VAS, but then I'm going to need Miller comp and it's starting to become a different amp....
 
You must log in or register to reply here.
Top Bottom