MN3005 vs. MN3205 Questions

[blueduck577]

From what I understand, these are very similar chips.  What's the advantage of using one over the other (besides availability)?  This is what I've gathered:

• MN3005 is designed to work off of a "negative" VDD, MN3205 is designed for "positive" (i know about the workaround)
• MN3005 VDD can be up to -16V, MN3205 can be up to 9V

Anything significant I've missed?  Also, it says on the MN3205 datasheet the minimum VDD +4V to +9V whereas the MN3005 is -14V and maximum -16V.  Is this really set in stone?  I've run my MN3007 off of 9V before no problem, even though the datasheet specifies a minimum of 14V.  I suppose the MN3205 would be more friendly to a 9V guitar pedal circuit?

Thanks guys.

[Mark Hammer]

As near as any of us can figure out, the operating voltage range of the MN3205 was selected such that it could operate off a regulated 5v supply derived from a 9v battery.  In other words, the battery feeds a 78L05 with 9v.  The 78L05 feeds the MN3205 with 5v.

Why is this "better"?  Because the bias voltage is normally set as a fraction of Vdd.  That means that as Vdd changes over the life of the battery, the bias voltage will too, rendering it potentially off, if only by a bit.  Using the regulator to make Vdd 5v, means that when you set the bias, it's set.  The battery would have to drop to below 7vdc for the regulator to stop functioning properly.  By the time the battery has dropped that low, there generally isn't enough juice left in it to power anything else in the circuit, so the regulation trick is valid for the viable life of the battery.  The lower MN3205 Vdd cutoff of 4V is probably just to account for individual variation in regulators.  I.E., if part-to-part variation meant that the 78L05 in some unit actually put out a stable 4.73v, the MN3205 would still work, and the bias voltage, even though different, would still be set properly....for that unit.

[blueduck577]

hm. that makes a lot of sense.  but if the MN3205 is running off of a 5V supply, I suppose that means the voltage swing is limited to ±2.5V... which brings up another related question...

how necessary is a compander in an analog delay circuit?  With only ±2.5V swing available, I'd imagine some guitars with hot pickups or a boosted signal will send the thing into clipping without the compander

[Mark Hammer]

Yes, companders are generally necessary in an analog delay.  The principle reason I would think is that when people use them, they generally play in such a way that repeats can be heard AS repeats...and that means periods of silence between them.  Were it the case that it was standard practice to use buffered clock lines, a bunch of BBDs, and a super-high-frequency clock, then the potential clock noise wouldn't be an issue.  But the fact of thematter is that, until very recently with Coolaudio chips, 4096-stage BBDs were so pricey that very few designs ever used more than one or two.  When you aim for 300ms or more using a single 4096-stage device, the needed clock frequency moves well into the audio range.  Consequently, companders have been de rigeur for most analog delays.  I would imagine that containing dynamic range is also part of the reason, but keep in mind that no matter how softly one might play, and no matter what one precedes the delay line with, that dang clock is still going to be there, so it has generally been the primary target of companding in delays.

When companding was not employed, one often saw an input attenuator control and level/clipping LED.  I have one or two such companderless analog delays.  Given that delays generally have no gain (other than whatever compensation for loss needs to be inserted to provide effect/bypass volume balance), the vast majority of normal nput signals will be manageable in the absence of companding.  But, like I say, input clipping is only one of your concerns when it comes to signal optimization in delays.

[aziltz]

oh oh oh.

[cpm]

hm. that makes a lot of sense.  but if the MN3205 is running off of a 5V supply, I suppose that means the voltage swing is limited to ±2.5V... which brings up another related question...

well, datasheet says input v is 0.3Vrms at 5v, for best THD performance, so you are moving around a small peak to peak voltage, noise prevention is critical.

[Mark Hammer]

Because of the BBD headroom limitations, many companderless designs for flangers will include a diode-based soft-clipping circuit at the point where the recycled signal is mixed with the input.  This keeps their sum from ever getting out of hand.  See the Boss BF-2 and the A/DA Flanger for examples of this.

Of course, in the case of analog delays, even maxed-out repeat settings won't necessarily result in anything being summed with something else that could challenge headroom.  That's the difference between a 1-8msec time lag between input and regen signal, versus a 50-300msec time lag between input and regen signal.  The sound quality still degrades with repeats, but more because of sample-holding limitations of the BBD, and pre-post filtering, than because of headroom limitations.