[Amps] Fw: diode junction temps and thermal resistance
Jim Thomson
jim.thom at telus.net
Wed Oct 20 13:41:41 PDT 2010
Subject: Re: [Amps] diode junction temps and thermal resistance
Jim, in using the string of diodes for bias, do you notice any drift in the idle current as they warm up? Purchased in quantity these big diodes are way less expensive than a big zener diode and would be easy to mount on a small perfboard.
Makes one wonder why some company doesn't offer premade zener substitutes along the line of K2AW's hv diode designs.
Is there a downside to this method of generating cathode bias?
73,
Gerald K5GW
### It does not budge on 7 x different amps! [idle zsac]. Henry radio used the eq of a K2AW block for bias for their
8K ultra. Inside the single module was 15-20 x diodes in series. I consistently measure a 10% increase
in V drop across each diode, when toggling between say 200 ma...and 2A. That was done on the test jig.
I was trying to simulate idle current / full bore CXR on a 3CX-3000A7. I toggled between 200 ma + 4A as well,
and the diff was 12%. On ssb, you won't see it vary much at all. Put a big lytic across the entire string, and it
won't even budge on ssb. [The more current current you pull... the bigger the V drop. If the diode gets hot, the V drop
will go down, so the 2 x sorta negate each other]
## before somebody goes off on a tangent abt Bias regulation, consider the big V drop across the grid chokes
in a SB-220. [ another reason I toss em and strap the grids to the chassis] In any circuit that uses electronic grid
protection schemes, they also have a V drop across a sense resistor. In both cases that's just more un- wanted bias.
## If you cut open a K2AW block...sometimes you will find a string of 1N5408's inside. If you mount the diodes on
perf board, and keep the leads long, then you can pull more average current through them. I don't even mess with
1N5408's these days. Just buy 6A10's in bulk. Both have the same 16 ga leads on each end..and both are the same length.
A 6A10 is double the diam of a 1N5408 [ in 1 kv]. The 1 kv variety are bigger diam then the 800 v types. You can get em
down to 50 v, but don't bother. the bigger diam 1 kv units handle heat better. I use the 1 kv type in HV supplies as well
as meter protection, and bias, etc. The 6A10 in 1 kv format....bought in bulk is a good one diode does everything.
Jim VE7RF
In a message dated 10/20/2010 9:52:59 A.M. Central Daylight Time, jim.thom at telus.net writes:
Date: Wed, 20 Oct 2010 08:08:08 +0100
From: Ian White GM3SEK <gm3sek at ifwtech.co.uk>
Subject: Re: [Amps] diode junction temps and thermal resistance
Bill, W6WRT wrote:
>If not overtemp, it's almost certainly arcing inside the tube. Even
>good tubes arc occasionally with no harm to the tube, but there can be
>harm to components in the HV- to HV+ path and that includes the bias
>zener(s).
>
>IMO, for a legal-limit amp a ten watt zener is marginally adequate.
>Even if it can dissipate the cathode current x zener voltage, there
>isn't much left to absorb arcs. I would recommend a 50 watt stud mount
>type.
>
>There is a certain amount of guessing in the rating because you never
>know how massive the arc is going to be.
## a 50 watt zener requires a 50 w heatsink. The problem I had
with 10 w zeners back in the 70's is..they are the 1st component to short
out come glitch time. The problem with both a 10 w and a 50 w zener's
is they won't handle a surge worth a damn. If you use a zener, use a
50W unit.
Actually you *can* know something about the arc.
1. The peak current. The main function of a surge limiting resistor in
the B+ line is to limit the peak current to a known value, so that Ipk =
Vb/R. The actual arc current will be time-varying but it can never
exceed that value of Ipk, so you *do* have a known value to design
against.
### use a 50 ohm glitch R in the B+ lead, instead of the usual
10-25 ohm unit, and now you can really limit the max peak current.
A 50 ohm glitch can consists of parallel 100 ohm WW's... or
2 x 25 ohm WW's in series. I use a single 50 ohm-50W WW
in my L4B's... zero problems. 2650 vdc / 50 ohms = 53A max fault
current..... then the B+ fuse blows open in < 2 msecs.
2. The time it will take to shut the HV supply down (by means of a fuse,
breaker, relay, crowbar or whatever means you have chosen). It isn't
easy to get beyond an initial guesstimate of "a few milliseconds", but
further detail *is* knowable if you're prepared to drill deep enough.
## I use a B+ fuse.....just before the glitch R. I also use a 2nd HV fuse,
between sec of plate xfmr and input of diode board [ one leg only]. Then
a magnetic hydraulic breaker in the 240 vac primary. Now you van cro-bar the
B+ to chassis all day long...and nothing happens....except a blown B+ fuse.
On the 7 kv supply, the B+ fuse will open off < 2 msecs.
## I use a FAST 3agc type cathode fuse, in the CT of the fil xfmr, with a
pair of 100k-3W MOF's across it. For a grid fuse, I used another FAST 3agc
type fuse, wired between chassis and neg terminal of grid meter [or grid shunt].
Nothing gets wired across the grid fuse. if grid fuse blows open, no path for dc grid
current, amp can't be driven, and you end up with zero power out of amp. Input
swr rises to infinity. Input wattmeter detects the high swr and shut's down the
XCVR..and latchs it off. [ the xcvr would shut itself down anyway].
The big unknown (to me, at least) is what fraction of the total arc
current will flow through the cathode zener or other bias circuit. To
trace the current flows, see Figure 2 of:
http://www.ifwtech.co.uk/g3sek/boards/triode/triode-manual.pdf
If the grid collects 100% of the arc current, then none of it will flow
through to the cathode. The current will be steered around that part of
the circuit, flowing through the B-minus clamping diodes instead, and
the cathode zener (or other bias circuit) will be unaffected.
## good point. Instead of a zener, on the hb amps I use a series
string of 6A10 diodes. Then with a 12 / 20 pos rotary switch, every
2nd or 3rd diode is tapped off.....starting at diode #10. For a
real fine adjust, a spst toggle is used across diode #1. To improve the
dynamic bias regulation, a big 10,000 to 200,000 uf lytic is wired across
the entire string of diodes. Now the bias can be adjusted on the fly. You can
suck 6A all day though it. With it's 400A surge rating, you can't blow it up anyway
so it never needs to be repaired, or replaced. No heat sink required either. If
you are real hardcore, parallel 6A10's work good too. [12A/800A surge]
## for simpler stuff, like my L4B.. I used 10 x 1N5408's in series..and a
SPDT-center off miniature toggle on the rear apron. That toggle gives me 3
positions of bias.
## If you have to replace zener's in a GG amp, they were not engineered
correctly to begin with...and ditto with safety diodes between B- and chassis,
and across plate + grid meter's.
Jim VE7RF
At the other extreme, if the arc punches clear through the grid
structure then a large fraction of the current will flow through the
cathode, causing damage to the cathode and possibly/probably the bias
circuit. For examples of cathode damage, see:
http://www.ifwtech.co.uk/g3sek/misc/gs35-arcs.gif
The GS35b is a planar triode with a disk-shaped oxide cathode and a mesh
("tea strainer") grid. Comparing the left-hand and right-hand images,
you can match up the marks on the anode with the corresponding marks on
the grid. If you then zoom into the left-hand image of the grid, you can
see where the largest arc has punched through to leave a burn mark on
the cathode underneath.
These kinds of damage would be much harder to see in other types of tube
construction, but I'm fairly sure that similar effects will exist.
The key to damage limitation is ALWAYS to install that surge limiting
resistor in the B+ line! Eimac Bulletin 17 explains exactly why:
http://www.ifwtech.co.uk/g3sek/misc/bull17.pdf
--
73 from Ian GM3SEK
http://www.ifwtech.co.uk/g3sek
_______________________________________________
Amps mailing list
Amps at contesting.com
http://lists.contesting.com/mailman/listinfo/amps
More information about the Amps
mailing list