Hi Group,

I have a 2500 that had one bad 8875, and as-is EBay buy, and I am in the
process of a conversion to a single 8877. This may not be the best choice of
a tube for the application, but I had a good one in a box for nearly 20
years, so I decided to give it a try. Before the conversion I powered it up,
and with one tube I could get 450 watts on 80M with 50 watts drive. I did
notice however that the variable loading capacitor had to be at full mesh on
any band I tried. I blew off that issue because with only one tube the
PI-Net was not seeing the as-designed input impedance.

My question is from my blurry schematic and parts list. The fixed
Centralab loading capacitors in my rig are all 1000pf, and the fixed plate
tuning side capacitors are all Centralab 100pf units. The parts list does
not show any 1000pf HV Centralab capacitors. Is there a typo in the parts
list, or does my amplifier have the wrong capacitors installed in the
loading area?

I powered up the amplifier last night, and I can get 600 watts RF out on
80 meters with 50 watts drive. The cathode current reaches 0.8 ampere at
1700 volts Plate. So that is 1360 watts in for 600 watts output. That is
about 44% efficiency. The variable loading control always wants to be at
maximum mesh for maximum RF output. On 160, 40, and 20m meters the loading
variable capacitor must also be at maximum mesh. This makes me think my
PI-Network is not correct for the impedances involved, and then it sure
appears that the Dentron documentation does not match my rig. I might have
to reverse engineer that PI-Net.

So does anyone have a better set of drawings to see if their MLA-2500 PI-Net
parts match the drawings?

Regards,
Jim
WD5JKO

Update:

The fixed loading capacitors on the MLA-2500B for 75, 160 meters are both
1000pf per Dentron parts list whereas they are 100pf on the MLA-2500 parts
list. It looks like a typo in the documentation. Then I measured my 1000pf
capacitors in my amp, these both measured about 820pf on my capacitance
bridge at 4 Mhz. For sanity sake, I also measured a 820 pf silver mica
capacitor, and the bridge read 850pf @ 4 Mhz. So it seems that I uncovered
two problems:

* Dentron MLA-2500 parts list has typos in it

* My two loading capacitors for 75 and 160 are out of tolerence on the low
side. This explains why my variable loading capacitor needs to be at full
mesh.

Anybody have a 8877 Teflon chimney they want to get rid of?

Jim
WD5JKO

"JC" wrote in message
. com...
Hi Group,

I have a 2500 that had one bad 8875, and as-is EBay buy, and I am in the
process of a conversion to a single 8877. This may not be the best choice
of a tube for the application, but I had a good one in a box for nearly 20
years, so I decided to give it a try. Before the conversion I powered it
up, and with one tube I could get 450 watts on 80M with 50 watts drive. I
did notice however that the variable loading capacitor had to be at full
mesh on any band I tried. I blew off that issue because with only one tube
the PI-Net was not seeing the as-designed input impedance.

My question is from my blurry schematic and parts list. The fixed
Centralab loading capacitors in my rig are all 1000pf, and the fixed plate
tuning side capacitors are all Centralab 100pf units. The parts list does
not show any 1000pf HV Centralab capacitors. Is there a typo in the parts
list, or does my amplifier have the wrong capacitors installed in the
loading area?

I powered up the amplifier last night, and I can get 600 watts RF out on
80 meters with 50 watts drive. The cathode current reaches 0.8 ampere at
1700 volts Plate. So that is 1360 watts in for 600 watts output. That is
about 44% efficiency. The variable loading control always wants to be at
maximum mesh for maximum RF output. On 160, 40, and 20m meters the loading
variable capacitor must also be at maximum mesh. This makes me think my
PI-Network is not correct for the impedances involved, and then it sure
appears that the Dentron documentation does not match my rig. I might have
to reverse engineer that PI-Net.

So does anyone have a better set of drawings to see if their MLA-2500
PI-Net parts match the drawings?

Regards,
Jim
WD5JKO

It is unlikely that the fixed loading capacitor is as high as 1000 pf, except for 160m. However if Dentron has really used a Centralab 1000-pf capacitor as fixed loading capacitor, that is a sign of bad design, because those capacitors are not suitable for that purpose.

As a matter of fact the Centralab 1000-pF capacitors use a different ceramic material than the 100 pF units, and they therefore have a very different behavior, as explained below:

- the 1000-pF units can withstand high RF current, but cannot withstand high RF voltage at the same time (DC voltage does not matter). A more rigorous way to express that concept is to say that the 1000-pf units cannot withstand high "reactive power" (reactive power = RF current times RF voltage). In practice they are only suitable for bypassing or plate-to-pi coupling (high RF current but very little RF voltage).

- on the contrary, the 100-pF units are suitable for operation in resonating circuits, where both the RF voltage and the RF current is high (= high reactive power). So they are suitable as fixed loading capacitors.

If one uses a Centralab 1000-pF capacitor as the fixed loading capacitor of a high-power linear, the capacitor will heat up and change its capacitance, so detuning the circuit. Prolonged high-power operation may result in capacitor failure.

73

Tony I0JX




..


"JC" ha scritto nel messaggio . com...
Hi Group,

I have a 2500 that had one bad 8875, and as-is EBay buy, and I am in the
process of a conversion to a single 8877. This may not be the best choice of
a tube for the application, but I had a good one in a box for nearly 20
years, so I decided to give it a try. Before the conversion I powered it up,
and with one tube I could get 450 watts on 80M with 50 watts drive. I did
notice however that the variable loading capacitor had to be at full mesh on
any band I tried. I blew off that issue because with only one tube the
PI-Net was not seeing the as-designed input impedance.

My question is from my blurry schematic and parts list. The fixed
Centralab loading capacitors in my rig are all 1000pf, and the fixed plate
tuning side capacitors are all Centralab 100pf units. The parts list does
not show any 1000pf HV Centralab capacitors. Is there a typo in the parts
list, or does my amplifier have the wrong capacitors installed in the
loading area?

I powered up the amplifier last night, and I can get 600 watts RF out on
80 meters with 50 watts drive. The cathode current reaches 0.8 ampere at
1700 volts Plate. So that is 1360 watts in for 600 watts output. That is
about 44% efficiency. The variable loading control always wants to be at
maximum mesh for maximum RF output. On 160, 40, and 20m meters the loading
variable capacitor must also be at maximum mesh. This makes me think my
PI-Network is not correct for the impedances involved, and then it sure
appears that the Dentron documentation does not match my rig. I might have
to reverse engineer that PI-Net.

So does anyone have a better set of drawings to see if their MLA-2500 PI-Net
parts match the drawings?

Regards,
Jim
WD5JKO

Tony,

I appreciate the comments. For sure there is something awry in this
pi-net. I looked at your tutorial on your web page, and wow, you know your
stuff!
The schematics of the 2500, and 2500B are at the following link:

ftp://bama.sbc.edu/downloads/dentron/mla2500/

Another aspect of this pi-net is that the tank coil consists of tight
wound silver plated stranded Teflon insulated wire of about 10 awg. Users of
healthy 2500's report 1100 to 1300 watts key down (output), and 1500 watts
PEP out with a pulser. This is very good for the size and weight of this
amplifier.

The loading variable cap is about 50-500pf with a 100pf Centralab in
parallel with that. The bandswitch is not a shorting type, so it adds 200pf
on 40m, 1000pf on 75, and 1000pf on 160 (all on 2500B schematic and parts
list). These do not add up due to the type of rotary switch used. I'd prefer
a shorting switch arrangement where on 160m, all the fixed loading caps are
in parallel and each unit has less capacitance.

I will look more carefully at the capacitors used to see what the part
number is. Maybe the correct values are in there, but of the wrong type. I
plan to use my impedance bridge from the 50 ohm amp output looking into the
amp, and replace the 8877 with the calculated plate load RL going into the
pi-net. Hopefully a 1/2 watt carbon comp resistor around 2K (a guess) will
represent the plate side impedance. My hope here is that I can evaluate the
pi-net 160-10m at a low level (0dbm?) to see what the impedance matching
capability is. Maybe the values used are correct, but with high power, the
capacitors get lossy (hot), drift down in value, and cause the efficiency of
the amp to really stink.

I'll let you'all know what I find.

Thanks,
Jim
WD5JKO



"Antonio Vernucci" wrote in message
...
It is unlikely that the fixed loading capacitor is as high as 1000 pf,
except for 160m. However if Dentron has really used a Centralab 1000-pf
capacitor as fixed loading capacitor, that is a sign of bad design, because
those capacitors are not suitable for that purpose.

As a matter of fact the Centralab 1000-pF capacitors use a different ceramic
material than the 100 pF units, and they therefore have a very different
behavior, as explained below:

- the 1000-pF units can withstand high RF current, but cannot withstand high
RF voltage at the same time (DC voltage does not matter). A more rigorous
way to express that concept is to say that the 1000-pf units cannot
withstand high "reactive power" (reactive power = RF current times RF
voltage). In practice they are only suitable for bypassing or plate-to-pi
coupling (high RF current but very little RF voltage).

- on the contrary, the 100-pF units are suitable for operation in resonating
circuits, where both the RF voltage and the RF current is high (= high
reactive power). So they are suitable as fixed loading capacitors.

If one uses a Centralab 1000-pF capacitor as the fixed loading capacitor of
a high-power linear, the capacitor will heat up and change its capacitance,
so detuning the circuit. Prolonged high-power operation may result in
capacitor failure.

73

Tony I0JX




..


"JC" ha scritto nel messaggio
. com...
Hi Group,

I have a 2500 that had one bad 8875, and as-is EBay buy, and I am in
the
process of a conversion to a single 8877. This may not be the best choice
of
a tube for the application, but I had a good one in a box for nearly 20
years, so I decided to give it a try. Before the conversion I powered it
up,
and with one tube I could get 450 watts on 80M with 50 watts drive. I did
notice however that the variable loading capacitor had to be at full mesh
on
any band I tried. I blew off that issue because with only one tube the
PI-Net was not seeing the as-designed input impedance.

My question is from my blurry schematic and parts list. The fixed
Centralab loading capacitors in my rig are all 1000pf, and the fixed plate
tuning side capacitors are all Centralab 100pf units. The parts list does
not show any 1000pf HV Centralab capacitors. Is there a typo in the parts
list, or does my amplifier have the wrong capacitors installed in the
loading area?

I powered up the amplifier last night, and I can get 600 watts RF out
on
80 meters with 50 watts drive. The cathode current reaches 0.8 ampere at
1700 volts Plate. So that is 1360 watts in for 600 watts output. That is
about 44% efficiency. The variable loading control always wants to be at
maximum mesh for maximum RF output. On 160, 40, and 20m meters the loading
variable capacitor must also be at maximum mesh. This makes me think my
PI-Network is not correct for the impedances involved, and then it sure
appears that the Dentron documentation does not match my rig. I might have
to reverse engineer that PI-Net.

So does anyone have a better set of drawings to see if their MLA-2500
PI-Net
parts match the drawings?

Regards,
Jim
WD5JKO

"JC" wrote in message
. com...
Hi Group,

I have a 2500 that had one bad 8875, and as-is EBay buy, and I am in the
process of a conversion to a single 8877. This may not be the best choice
of a tube for the application, but I had a good one in a box for nearly 20
years, so I decided to give it a try. Before the conversion I powered it
up, and with one tube I could get 450 watts on 80M with 50 watts drive. I
did notice however that the variable loading capacitor had to be at full
mesh on any band I tried. I blew off that issue because with only one tube
the PI-Net was not seeing the as-designed input impedance.

My question is from my blurry schematic and parts list. The fixed
Centralab loading capacitors in my rig are all 1000pf, and the fixed plate
tuning side capacitors are all Centralab 100pf units. The parts list does
not show any 1000pf HV Centralab capacitors. Is there a typo in the parts
list, or does my amplifier have the wrong capacitors installed in the
loading area?

I powered up the amplifier last night, and I can get 600 watts RF out on
80 meters with 50 watts drive. The cathode current reaches 0.8 ampere at
1700 volts Plate. So that is 1360 watts in for 600 watts output. That is
about 44% efficiency. The variable loading control always wants to be at
maximum mesh for maximum RF output. On 160, 40, and 20m meters the loading
variable capacitor must also be at maximum mesh. This makes me think my
PI-Network is not correct for the impedances involved, and then it sure
appears that the Dentron documentation does not match my rig. I might have
to reverse engineer that PI-Net.

So does anyone have a better set of drawings to see if their MLA-2500
PI-Net parts match the drawings?

Regards,
Jim
WD5JKO
Jim -

For reference purposes, here are the Communication Quarterly and KU7G
articles (Adobe Acrobat)
regarding 4CX400a or 4CX800a conversions
http://www.tomstubes.com/images/mla_2500_4cx400a.pdf

KU7G
http://www.tomstubes.com/images/mla2500_4cx800a.pdf

w9gb

For reference purposes, here are the Communication Quarterly and KU7G
articles (Adobe Acrobat)
regarding 4CX400a or 4CX800a conversions
http://www.tomstubes.com/images/mla_2500_4cx400a.pdf

KU7G
http://www.tomstubes.com/images/mla2500_4cx800a.pdf

w9gb


Hi, Thanks for the excellent links. I had seen the first one before. In
looking at the article about the 2 times 4cx400a, it jumps out at me that
this conversion leads to about 350-500 watts less RF output than original
(1100-1300 versus 750-800) even when using tubes with more capability. I
attribute this to running Tetrodes at low plate voltage. In this application
Hi-Mu Triodes are a better choice. Also the low RF drive need for the
Tetrodes make them attractive for low power exciters, and a problem for us
with a 100 watt exciter.

My conversion is similar to the 4cx800 article except I am using a 8877. I
use the same external box to house the socket, and mount the fan. The 8877
may not be designed for use at 1700 volts, but what the heck, I only want
about 1000 watt RF output. I need a chimney like what is described for tthe
PI-Network he 4cx800. Any idea where I can get a block of Teflon, and then
get it machined?

That article about the 4cx800 has some good visuals of the pi-net that is
being a problem with my 2500.

Thanks again,

Jim
WD5JKO

"JC" wrote in message
et...


For reference purposes, here are the Communication Quarterly and KU7G
articles (Adobe Acrobat)
regarding 4CX400a or 4CX800a conversions
http://www.tomstubes.com/images/mla_2500_4cx400a.pdf

KU7G
http://www.tomstubes.com/images/mla2500_4cx800a.pdf

w9gb


Hi, Thanks for the excellent links. I had seen the first one before. In
looking at the article about the 2 times 4cx400a, it jumps out at me that
this conversion leads to about 350-500 watts less RF output than original
(1100-1300 versus 750-800) even when using tubes with more capability. I
attribute this to running Tetrodes at low plate voltage. In this
application Hi-Mu Triodes are a better choice. Also the low RF drive need
for the Tetrodes make them attractive for low power exciters, and a
problem for us with a 100 watt exciter.

My conversion is similar to the 4cx800 article except I am using a 8877. I
use the same external box to house the socket, and mount the fan. The 8877
may not be designed for use at 1700 volts, but what the heck, I only want
about 1000 watt RF output. I need a chimney like what is described for
tthe PI-Network he 4cx800. Any idea where I can get a block of Teflon, and
then get it machined?

That article about the 4cx800 has some good visuals of the pi-net that is
being a problem with my 2500.

Thanks again,

Jim
WD5JKO
Jim,

Bob Schetgen, KU7G machined that Teflon block in the 4CX800 article .
Unfortunately, Bob passed away on December 5, 2005.
Bob worked at the ARRL Headquarters for nearly 23 years.
He was a member of the editorial staff for the past 16 years, he was perhaps
best known as the editor of the "Hints & Kinks" column in QST.

He also served as the managing editor of the League's technical and
experimenter's journal QEX.

w9gb

On Sun, 21 May 2006 09:16:28 -0500, "g. beat" @spam protected
wrote:

"JC" wrote in message
.com...
Hi Group,

snip
So does anyone have a better set of drawings to see if their MLA-2500
PI-Net parts match the drawings?

Regards,
Jim
WD5JKO
Jim -

For reference purposes, here are the Communication Quarterly and KU7G
articles (Adobe Acrobat)
regarding 4CX400a or 4CX800a conversions
http://www.tomstubes.com/images/mla_2500_4cx400a.pdf

KU7G
http://www.tomstubes.com/images/mla2500_4cx800a.pdf

I would suggest an alternate chimney though using Silastic rubber
sheet wound into a tube shape as is done in the Alpha series amps.
Relatively inexpensive, no machining, and easy to install.

One other comment. The MLA 2500 suffers from poor voltage regulation
as did most of the Dentron amps. It does after all use a doubler in
the power supply. So when tuning be sure to use the "tune" position
and then switch to operate. Tuning in the operate position drags the
plate voltage down so far that the amp will not be properly tuned for
the PEP on SSB. With my voice characteristics and under the old rules
of 1 KW average input mine would put out about 2200 PEP.

Good Luck,

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com


Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com

w9gb

On Sun, 21 May 2006 15:50:55 GMT, "JC" wrote:



For reference purposes, here are the Communication Quarterly and KU7G
articles (Adobe Acrobat)
regarding 4CX400a or 4CX800a conversions
http://www.tomstubes.com/images/mla_2500_4cx400a.pdf

KU7G
http://www.tomstubes.com/images/mla2500_4cx800a.pdf

w9gb


Hi, Thanks for the excellent links. I had seen the first one before. In
looking at the article about the 2 times 4cx400a, it jumps out at me that
this conversion leads to about 350-500 watts less RF output than original
(1100-1300 versus 750-800) even when using tubes with more capability. I
attribute this to running Tetrodes at low plate voltage. In this application
Hi-Mu Triodes are a better choice. Also the low RF drive need for the
Tetrodes make them attractive for low power exciters, and a problem for us
with a 100 watt exciter.

My conversion is similar to the 4cx800 article except I am using a 8877. I
use the same external box to house the socket, and mount the fan. The 8877
may not be designed for use at 1700 volts, but what the heck, I only want

As I recall the MLA-2500 ran 2200 volts.

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com

about 1000 watt RF output. I need a chimney like what is described for tthe
PI-Network he 4cx800. Any idea where I can get a block of Teflon, and then
get it machined?

That article about the 4cx800 has some good visuals of the pi-net that is
being a problem with my 2500.

Thanks again,

Jim
WD5JKO

"Roger" wrote in message
...
On Sun, 21 May 2006 09:16:28 -0500, "g. beat" @spam protected
wrote:


I would suggest an alternate chimney though using Silastic rubber
sheet wound into a tube shape as is done in the Alpha series amps.
Relatively inexpensive, no machining, and easy to install.

You can purchase the Silastic rubber from Alpha directly.
It is no longer red, but black -- supplier to Alpha changed this !

w9gb


Good Luck,

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com


Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com