The Newton 300B

[Thermionic Idler]

OK so some tweaking required later this week. I have a suspicion that the current consumption of both amps together may be above what the variac is rated for - certainly the turn on surge will be, so we'll just have to live with things as is until I get a chance to go under the bonnet again.

I need to re-measure the heater voltages on both boxes along with the mains voltage so I get an idea of what dropper resistors are needed, and whether the mains varies beyond the 245V I measured this morning. They definitely need to come down, long term those values are not good for valve lifespan especially as these are being run quite hard anyway.

Whilst I have them on the bench, I think I'm also going to parallel the 500R shared 300B cathode resistor with a 5 watt 4.7k, as the plate current (77mA) is a little outside Lynn Olson's recommended 80-85mA. I think this is because I changed to using a GZ34 recifier, which has more sag than the TV damper diodes in the original sketch, so my HT is dropped a bit from the predicted 500V. According to my Spice model, that should get me back into the middle of the range current-wise (about 83mA). The B+ will fall a little to compensate.

Hopefully I'll get a chance to do that this coming Friday.

[Thermionic Idler]

Well, I re-measured all the heater voltages on Thursday morning. On both amps:

12SN7: 14.1 ( = 7.05 across both 6V6's)
5AR4 output stage: 5.3
5AR4 input stage: 5.55
300B: 5.48.
Voltage on mains: 245V.

I ordered a set of 5% resistors yesterday to drop the input 5AR4 voltage to about 5.3, which brings it well within the +/-10% tolerance specified on the datasheet, and about the same value seen on the other rectifier. Considering both 5V secondaries on the Hammond 375X and 378X are rated the same, seems odd that I get such a significant difference.

Then some 1% resistors to drop the signal valve heater voltages to their exact required levels. These actually fell nicely into place - the input stage valves neeeded a Vdrop of 1.5V against a total heater current of 750mA - that'll be two 1 Ohm resistors then. Same with the 300B's - Vdrop 0.48V, heater current 2400mA for the two valves - a nice round 250 mOhm.

I had planned to make the mods tomorrow morning, as the resistors were despatched from RS yesterday for next day delivery. Unfortunately, Parcelforce missed their SLA and didn't deliver today as expected so I'm probably going to have to shoehorn this in on Monday unless they deliver sometime tomorrow.

[Ray P]

IIRC, Parcelforce don't deliver on Saturdays - all part of the service...

[Thermionic Idler]

IIRC, Parcelforce don't deliver on Saturdays - all part of the service...

Yeah the message on the website, when I went in to the parcel tracker, said I wouldn't see it until Monday, along with the usual non-apology (I don't view automatically generated "we are sorry" text as genuine contrition).

EDIT - gotta love Parcelforce. For some reason unknown to me, RS decided to split the order into two packages even though everything was in stock and they were despatching the same day. I imagine both deliveries had the same SLA.

Today (Saturday) I received one of those packages, but not the other one! Checking the tracker confirmed the guy didn't just leave it in his van - one went out for delivery, the other stayed at the depot. Brilliant.

The good thing is that the delivery I just received were the 0.2 ohm dropper resistors for the 300B, and I also have the Mills 4.7k from HFC to drop the cathode R. I think I'll make that mod this weekend, because with the replacement cost of 300B's, I don't want them to be running over-voltage for longer than is absolutely necessary.

[Thermionic Idler]

Well, yesterday was fun.

I decided to at least sort out the 300B operating conditions as I had the resistors to do that. A simple matter of adding a 0.2 ohm dropper into the heater line and parallelling the 4.7k on the shared cathode.

This went pretty smoothly on the first amp, although getting the dropper resistor in at the filament transformer was a bit awkward. The resulting heater voltage - 4.92V - was far more in line with what it should be. The cathode resistance went to a measured 452 ohms. The resulting plate voltage was measured at about 385V, with 75.1V bias. So about 83mA across each 300B.

It was on the second amplifier that everything seemed to go wrong. First I had the pain-in-the-ass job of soldering in the dropper resistor - then when I tested the heater circuit, there was no heater voltage at all. Turned out the dropper resistor was faulty and had gone open-circuit, so I had to dig it back out again and solder in another one (thankfully I got a packet of 5). After that, all was well with 4.92V again measured on the heaters.

After adding the cathode R, I powered up the amp to measure the positive phase plate voltage - 378V. Powered down the amp and moved the clip leads to the other valve. Waited a good minute or two then powered back up to measure the other plate voltage. Hmm, what's that growling noise? Then I noticed the output stage rectifier was arcing quite impressively, at the same time the plate voltage dropped like a stone - it had blown one of the HT fuses and the resistance on the other fuse was high so that was probably just about to blow before I powered it off. I wondered if the output stage was pulling too much current, however the previous measurement I'd made didn't seem to indicate any issue - I'd have expected that plate voltage to be a lot lower. I tried swapping the rectifiers over, reasoning that the input stage pulls a lot less current, but the same thing happened, and it took out both HT fuses on the input stage.

So I then powered down, took out the arcing rectifier, replaced the output stage fuses, and powered back up. Then the second GZ34 starting arcing, and I seemingly have two dud GZ34's. Getting rather worried at this point. I replace the output stage fuses again, then crossing my fingers "third time lucky" I went and got one of the rectifiers from the amp I'd successfully tested earlier in the day. This time all was well, the same plate voltage was measured on the negative phase with 76.9V bias. So here, 84.9mA across each 300B, with the plate dissipation working out to be about the same - 32 watts.

I had no spare GZ34's but I did have some Mullard GZ32's - I think I got them years back as spares when I had a pair of Quad II's. The input stage circuit pulls no more than 92mA and the data sheet I looked at gave a max of 125mA at 500-0-500 with cap input, so I subbed those into the input circuit, and we had working amps again!

The sound has definitely improved as a result of the changes. Before, some program material would reveal a bit of hardness or stridency in the upper mids, this seems to have gone away completely now, and the overall tone is a little rounder with improved soundstage depth. Also the bass registers seem to have tightened up a little bit too. So things are going in the right direction

I'm a bit mystified about why I had two GZ34 failures in quick succession though. The current draw has only increased by 12-14mA on the output stage to about 176mA, and that is still well within the design maximum of 250mA for the GZ34. Maybe they were marginal and the increase was enough to tip them over the edge. I have read that powering up and down in quick succession can cause rectifiers to arc, and maybe I wasn't leaving a long enough interval between cycles - I tried to give it a reasonable time, a good couple of minutes or so.

[pre65]

Re the GZ rectifiers, have you got enough resistance in the transformer ?

From memory last time I used GZ32/34 an extra 72R per anode was needed.

EDIT

Rt = 2 x 125R minimum.

Also, the first capacitor has a max value. The datasheet I looked at said 60uf, but I had a feeling it was less than that.

[Cressy Snr]

Yes,
I can’t see any series current limiting resistors on the schematic. New production gz34s are not as robust as the old ones.
Don’t get me or Simon started on JJs
Something like a 150R resistor before the first cap is a good alternative to resistors in each leg. That’s what I’ve used on my amp’s power supplies.

[Nick]

Where were the HT fuses? Before or after the rectifiers?

[Thermionic Idler]

Where were the HT fuses? Before or after the rectifiers?

Before the rectifiers.

[Thermionic Idler]

OK, some data.

Here's the power supply in question:



The datasheet for the Hammond 378X states the DCR between the two 400V connections to be 92.81 ohms, which at first glance doesn't seem like enough.

However, the datasheet for the older GZ32 gives varying Rt depending on the value of the first capacitor - it requires 150R for 60uF but only 50R for 16uF, so presumably the required resistance reduces further if the first cap seen is only 5uF, and could be covered by the transformer's DCR?

The GZ34 datasheet gives values for only 60uF but of course my first cap is a fraction of that value, and is followed by a choke.

EDIT - on the smaller input stage transformer, the 400V - 400V DCR is 135.1 ohm.

[izzy wizzy]

I find this helpful when choosing rectifiers https://www.hammfg.com/electronics/tran ... /rectifier

If a fuse blows, the rectifier then needs to be rated at twice what it was when both fuses are in circuit which might overload the rectifier.

[Thermionic Idler]

Hmm - so thinking about this further, the value of the first cap doesn't make any difference to the peak switching current, only the amount of time that peak lasts. It's interesting that they've given a sliding scale on the GZ32 datasheet for Rt (which suggests the GZ32 is actually fine on the input stage) but a blanket value for the GZ34.

It looks like I need to add a shade over 100R of resistance to the output stage secondary. Question is, how much will that upset my carefully curated 300B operating point! On the other hand, the 5U4, whilst having more sag, can cope with a lot less less plate supply resistance.

[Nick]

Remember that the GZ32mis aimed more at choke input supplies, hence the reduced first cap value in the datasheets.

A IN4007 in series with each rectifier anode will protect the transformer secondary in the case of an arcing rectifier. That combined with a single fuse in the centre tap would be how I would do it.

[Thermionic Idler]

Thanks Nick, that's useful to know. And thanks to Phil and Steve too, without your input I wouldn't have identified this.

So it's looking like series resistors are the way to go - a quick play in PSUD suggested I'd still end up dropping more voltage by going to a 5U4G alone, than by adding series resistors and sticking with the GZ34/5AR4. Plus I've already etched the chassis with "5AR4" by the rectifier sockets! The formula here suggests my Rlim is currently 55.8 ohms on the output stage, well under the 125 ohm per anode required, so that's definitely the root cause of the issue right there. Thankfully the current 5AR4's are holding for now, it seems frequent power cycling sends them over the edge.

The fall in voltage looks like it'll take about 3mA off the 300B plate currents, which I can live with. I can always tweak it back up again, plus it'll de-stress the output valves a tiny bit too.

It'll be easy to implement the resistors at least - the purple wires here connect the ends of the fuses to the rectifier plates, I can just replace those wires with, probably 75 ohm resistors in the output stage to give me a bit of margin. I still need to do the calculation for the input stage to find out what's needed there.

Sigh, another order from RS beckons...

[Thermionic Idler]

I've run the numbers and here are the resistors I need to order (I'd got the numbers for Rlim wrong in my previous post). Please feel free to check my working and point out if I have anything wrong...