Another question about core sense amplifiers

Brent Hilpert hilpert at
Fri Mar 13 15:30:40 CDT 2015

On 2015-Mar-12, at 6:45 AM, Noel Chiappa wrote:
>> From: Brent Hilpert
>> Similarly, the address-line drivers use a center-tapped primary
>> configuration as a cheap way of providing the bi-directional drive
>> current for the address wires (like push-pull audio).
> On re-reading this, to ensure that I had fully extracted the content into my
> brain, I realized I didn't fully grok this. Could you expand a tiny bit on
> this (and especially the push-pull reference)?
> On thinking about it, I guess that what's happening is that there are two
> driver transistors, each attached to the center tap and one end, but with the
> polarity reversed. A positive pulse through one produces a positive pulse on
> the output secondary, whereas a positive pulse through the other produces a
> negative pulse on the output secondary. Or am I mis-understanding?

That's not quite the circuit arrangement, but yes, that is the operating principle.
The center-tap is driven by the address-selection to +supply, one end is driven by the read-pulse driver for one polarity of pulse, the other end by the write-pulse driver for the other polarity, the latter both going to GND.

> What I don't get is why that's better than simply attaching two opposed
> transistors directly to the address lines, as one sees in the output stages
> of audio amplifiers, to handle the two halves of a sine wave.

In a direct-coupled arrangement a regulated dual-polarity power supply is needed, rather than single-polarity, and there's going to be some asymmetry and additional complexity to driving the output drivers in either a complementary or stacked-pair arrangement. With the transformers, the +/– pulse drivers are identical in design and components - there is no asymmetry - there's some benefit to that in easily getting matching characteristics of the +/– address pulses.
Even in that other design I was describing (from the late-70s), the transformers did the level-shifting to drive the output drivers as half are on the +supply and half on the –supply, which then directly drove the address wires.
For a single-polarity supply it could be done with an H-bridge-style arrangement, requiring more transistors.

There's lots of ways of doing it, with changing economies and tradeoffs. It's a mid-60s calculator, and while the design issues are not entirely the same, it's from the same period when every cheap 6- or 7-transistor radio had two audio transformers to implement push-pull drive - which is to say little transformers were a functional and economic option for the time. We don't bother today of course, as a zillion transistors in an IC are better and cheaper.

(Just for some additional context, it wasn't till the late-60s that solid-state audio really started focussing on direct-coupled designs and getting rid of the output transformer.)

More information about the cctalk mailing list